Effect of Interlayer Bonding on Superlubric Sliding of Graphene Contacts: A Machine-Learning Potential Study

Area of Science:

• Materials Science
• Tribology
• Computational Physics

Background:

• Superlubric sliding in layered materials is sensitive to surface defects.
• Atomistic understanding is hindered by computational costs and lack of accurate classical force fields for defected systems.

Purpose of the Study:

• To develop a reliable computational method for studying defected layered materials.
• To investigate the impact of interlayer bonding on the friction of bilayer graphene.

Main Methods:

• Developed a machine-learning potential (MLP) using graph neural networks.
• Trained the MLP against density functional theory calculations with iterative configuration space exploration.
• Simulated friction in aligned and incommensurate bilayer graphene interfaces.

Main Results:

• Interlayer bonding mildly affects aligned graphene interfaces.
• Friction coefficients significantly increase in incommensurate graphene interfaces due to interlayer bonding.
• The superlubric regime is nearly lost in defected incommensurate interfaces.

Conclusions:

• Machine-learning potentials offer an efficient approach for simulating defected layered materials.
• Interlayer bonding is a critical factor influencing superlubricity in incommensurate graphene.
• The developed methodology is adaptable to other defected layered material systems.