Application of Machine Learning in Predicting the Properties of Two-Dimensional Semiconductor Materials.

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Chemistry

Background:

• Next-generation electronics require advanced functional materials.
• Two-dimensional (2D) semiconductors offer tunable properties for nanoelectronics and optoelectronics.
• Traditional methods like Density Functional Theory (DFT) are computationally expensive for large-scale screening.

Purpose of the Study:

• To review recent advances in machine learning (ML)-driven property prediction for 2D semiconductors.
• To compare traditional computational simulations with ML algorithms for materials discovery.
• To outline future directions for ML in 2D semiconductor research.

Main Methods:

• Systematic review of ML applications in predicting 2D semiconductor properties.
• Comparison of classical ML algorithms (e.g., random forests) with deep learning models (e.g., graph neural networks) for bandgap prediction.
• Analysis of feature engineering strategies for magnetic property prediction.

Main Results:

• Deep learning models outperform classical ML in low-data regimes and complex material systems for bandgap prediction.
• Feature engineering significantly impacts accuracy and efficiency in magnetic property prediction.
• ML accelerates the discovery and prediction of critical material characteristics.

Conclusions:

• ML offers an efficient alternative to traditional methods for 2D semiconductor property prediction.
• Future research should focus on standardized databases, physics-informed ML, and multimodal modeling.
• Integrating ML with experimental and theoretical methods is crucial for overcoming current challenges.