Exploring Complex Reaction Networks Using Neural Network-Based Molecular Dynamics Simulation.

Area of Science:

• Computational Chemistry
• Materials Science
• Chemical Dynamics

Background:

• Ab initio molecular dynamics (AIMD) is crucial for studying reactive systems.
• High computational costs limit AIMD's application to longer timescales and larger systems.

Purpose of the Study:

• To develop a novel, computationally efficient method for investigating complex reaction networks.
• To explore reaction mechanisms of novel high explosives without predefined reaction coordinates.

Main Methods:

• Utilized molecular dynamics simulations powered by a neural network potential.
• Trained the neural network potential using a workflow combining AIMD and interactive molecular dynamics in virtual reality.
• Applied the method to study the decomposition of a novel high explosive (ICM-102).

Main Results:

• Achieved panoramic visualization of complex reaction networks for ICM-102 decomposition.
• Discovered new reaction pathways not easily identifiable with traditional methods.
• Demonstrated accelerated sampling of rare reactive processes.

Conclusions:

• Neural network-based molecular dynamics offers a powerful approach for exploring complex reaction mechanisms.
• This method pushes the boundaries of theoretical and computational chemistry towards experimental realism.
• The findings are significant for understanding chemical dynamics under extreme conditions.