A general-purpose machine-learning force field for bulk and nanostructured phosphorus.

Area of Science:

• Materials Science
• Computational Chemistry
• Condensed Matter Physics

Background:

• Elemental phosphorus exhibits complex behavior, making atomistic simulations challenging.
• Accurate simulation methods are crucial for understanding phosphorus allotropes and their applications.

Purpose of the Study:

• To develop a universally applicable force field for elemental phosphorus using machine learning.
• To enable accurate and efficient atomistic simulations of phosphorus-based materials.

Main Methods:

• Machine learning (ML) model trained on quantum-mechanical data (DFT+MBD).
• Validation through simulations of black and violet phosphorus exfoliation.
• Testing transferability across different phases (molecular to network liquid).

Main Results:

• A robust ML-driven force field for elemental phosphorus was successfully created.
• Accurate prediction of phosphorene and hittorfene monolayer formation.
• Demonstrated transferability across diverse phosphorus phases and structures.

Conclusions:

• ML-based force fields offer a powerful approach for simulating complex materials like phosphorus.
• This methodology facilitates next-generation materials modeling, particularly for layered solids.
• Potential for new insights into phosphorus and related materials in chemistry and physics.