Accelerated First-Principles Exploration of Structure and Reactivity in Graphene Oxide.

Area of Science:

• Materials Science
• Computational Chemistry
• Nanotechnology

Background:

• Graphene oxide (GO) is a widely studied material, but its atomic-scale structure is not fully understood.
• Understanding GO's structure is crucial for optimizing its properties and applications.

Purpose of the Study:

• To develop and apply advanced machine learning methods for rapid exploration of graphene oxide's chemical and configurational space.
• To elucidate the atomistic mechanisms of GO's thermal reduction.

Main Methods:

• Utilizing on-the-fly acceleration for first-principles molecular dynamics combined with message-passing neural network potentials.
• Employing machine learning for rapid sampling of chemical structures with minimal prior knowledge.
• Developing a realistic ten-nanometre scale structural model for GO thermal reduction simulations.

Main Results:

• The machine learning approach enabled rapid exploration of GO's structural landscape.
• Simulations accurately reproduced experimental findings, including X-ray photoelectron spectroscopy (XPS) data.
• Atomistic and mechanistic details of GO thermal reduction were rationalized.

Conclusions:

• The developed machine learning platform provides a robust method for routine, accurate, and predictive simulations of graphene oxide and other carbonaceous materials.
• This approach significantly advances the understanding of nanoscale material structures and transformations.