Fast-Tracking Transition-State Localization via Reaction Directional Analysis

Area of Science:

• Theoretical Chemistry
• Computational Chemistry
• Chemical Dynamics

Background:

• Transition-state localization is crucial for understanding chemical reaction mechanisms.
• Identifying transition states is computationally intensive and a significant challenge in theoretical chemistry.

Purpose of the Study:

• To introduce a novel, efficient, and reliable method for transition state searching.
• To overcome the computational demands associated with traditional transition state localization techniques.

Main Methods:

• Integration of Reaction Directional Analysis (RDA) with the dimer method to create RDA-D.
• RDA generates quasi-transition-state structures from initial and final state geometries using dynamic interpolation, structural optimization, and directional analysis.
• The dimer method refines these quasi-transition-state structures.

Main Results:

• RDA-D demonstrated an average speedup of 5.83 times over the Nudged Elastic Band (NEB) method in CPU time.
• The method reduced the number of required gradient evaluations by a factor of 4.74.
• Benchmark tests were conducted on various gas-phase and surface catalytic reactions.

Conclusions:

• RDA-D provides a robust, scalable, and automation-friendly framework for transition-state localization.
• The method eliminates the need for predefined reaction coordinates or chemically intuitive initial guesses.
• This advancement significantly enhances the efficiency of elucidating chemical reaction mechanisms.