Exploring Two-Dimensional Materials Thermodynamic Stability via Machine Learning.

Area of Science:

• Materials Science
• Computational Materials Science
• Machine Learning in Materials Discovery

Background:

• Despite growing interest, two-dimensional (2D) materials beyond graphene and transition metal dichalcogenides remain underexplored for diverse applications.
• Identifying novel 2D materials with desirable properties is crucial for expanding their practical use.
• Thermodynamic stability is a fundamental prerequisite for any material's application.

Purpose of the Study:

• To develop a machine learning model for predicting the thermodynamic stability of novel two-dimensional (2D) materials.
• To accelerate the discovery of new 2D materials with potential for various applications.
• To screen stable 2D materials for suitability in photoelectrocatalytic water splitting.

Main Methods:

• Utilized machine learning techniques to predict material stability based on formation energy and energy above the convex hull.
• Classified material stability into low, medium, and high categories.
• Evaluated novel 2D compounds using only composition and structural symmetry, without requiring atomic position information.

Main Results:

• Generated over a thousand novel 2D material candidates.
• Validated the model's stability classifications for five new materials using density functional theory (DFT) calculations.
• Identified Sn2SeTe and PbTe as promising candidates for photoelectrocatalytic water splitting, with Sn2SeTe being a novel discovery.

Conclusions:

• The machine learning approach effectively identifies thermodynamically stable 2D materials, facilitating the discovery of novel compounds.
• The model's ability to predict stability using basic material properties streamlines the search for promising candidates.
• The identified stable materials, particularly Sn2SeTe, show potential for applications like photoelectrocatalytic water splitting, opening new research avenues.