Automated Discovery and Refinement of Reactive Molecular Dynamics Pathways

Area of Science:

• Computational Chemistry
• Chemical Physics
• Materials Science

Background:

• Molecular dynamics (MD) simulations are crucial for understanding chemical reactions at the atomic level.
• Identifying reaction pathways and mechanisms within complex MD simulations remains a significant challenge.
• Existing methods for reaction path analysis often rely solely on initial and final states, potentially missing dynamic events.

Purpose of the Study:

• To present a novel energy refinement method, "nebterpolation," for identifying and characterizing reaction events in molecular dynamics (MD) simulations.
• To demonstrate the broad applicability of nebterpolation to complex, multi-molecular ab initio simulations.
• To improve the accuracy and completeness of reaction mechanism discovery in computational chemistry.

Main Methods:

• Developed "nebterpolation," an energy refinement technique that smooths reactive MD trajectories in internal coordinates.
• Applied nebterpolation to ab initio nanoreactor simulations involving hundreds of atoms.
• Initiated reaction path searches on the potential energy surface using smoothed trajectory data.

Main Results:

• Nebterpolation successfully identified and characterized complex reaction events in ab initio simulations.
• The method revealed a C-C coupling pathway for glycolaldehyde synthesis within a nanoreactor simulation.
• Minimum energy paths derived using nebterpolation were distinct from those obtained by methods relying only on endpoint information.

Conclusions:

• Nebterpolation offers a flexible and broadly applicable approach to analyzing reaction dynamics in MD simulations.
• The method enhances the discovery of novel molecules and reaction mechanisms.
• Nebterpolation is a key component of a new simulation paradigm for reaction mechanism discovery.