Multi-Level Protocol for Mechanistic Reaction Studies Using Semi-Local Fitted Potential Energy Surfaces.

Area of Science:

• Computational Chemistry
• Theoretical Chemistry
• Chemical Dynamics

Background:

• Investigating chemical reaction mechanisms is crucial for understanding chemical processes.
• Accurate theoretical studies often require significant computational resources.
• Developing efficient protocols for mechanism studies is an ongoing challenge.

Purpose of the Study:

• To propose a novel multi-level protocol for routine theoretical studies of chemical reaction mechanisms.
• To develop a computationally economical method for constructing reactive potential energy surfaces.
• To enable automated and routine mechanistic reaction studies.

Main Methods:

• Utilizing the Nudged Elastic Band (NEB) method for initial reaction path sampling.
• Employing machine learning (sGDML) to fit forces from accurate electronic structure calculations.
• Constructing semi-local reactive potential energy surfaces (PES) for reaction pathways.

Main Results:

• Successfully applied the protocol to unimolecular (Bergman cyclization) and bimolecular (S_N2) reactions.
• Achieved qualitative agreement for stationary-point geometries, reaction coordinates, and energy barriers with limited accurate calculations.
• Demonstrated qualitative agreement in vibrational frequencies and reaction rate coefficients.

Conclusions:

• The multi-level protocol significantly reduces computational cost while maintaining accuracy.
• The method is versatile, applicable to various reaction types, and can be readily automated.
• This approach facilitates routine theoretical studies of complex chemical reaction mechanisms.