Predicting and Understanding Work Functions of Double Transition Metal MXenes via Interpretable Machine Learning...

Area of Science:

• Materials Science
• Computational Chemistry
• Condensed Matter Physics

Background:

• MXenes are a promising class of 2D materials with tunable electronic properties.
• Understanding and predicting the work function of MXenes is crucial for their application in electronic devices.
• Double transition metal MXenes offer expanded possibilities for property tuning.

Purpose of the Study:

• To develop interpretable machine learning models for predicting the work functions of double transition metal MXenes.
• To identify key elemental features governing the work function of these materials.
• To provide a framework for the rational design of MXene-based materials with desired work functions.

Main Methods:

• First-principles calculations were used to generate a dataset of 242 double transition metal MXene structures.
• Various machine learning regression models, including Random Forest, were trained and evaluated.
• The Sure Independence Screening and Sparsifying Operator (SISSO) method was employed to derive analytical models.

Main Results:

• The Random Forest model demonstrated high predictive accuracy (R^2 = 0.86 ± 0.03).
• Feature importance analysis revealed that the outer transition metal (M1) has the dominant effect on work function.
• SISSO-derived analytical expressions achieved comparable accuracy (R^2 = 0.82 ± 0.04) and provided physical insights.

Conclusions:

• Interpretable machine learning provides an effective approach for predicting MXene work functions.
• The work function of double transition metal MXenes is primarily determined by the outer transition metal.
• These findings facilitate the accelerated discovery and design of novel MXene materials for specific applications.