Nudged Elastic Band Method for Molecular Reactions Using Energy-Weighted Springs Combined with Eigenvector Following

Area of Science:

• Computational Chemistry
• Chemical Reaction Dynamics

Background:

• The climbing image nudged elastic band (CI-NEB) method is crucial for identifying reaction coordinates and transition states.
• Standard CI-NEB implementations distribute images evenly, which can limit resolution in flexible systems.

Purpose of the Study:

• To enhance the resolution of CI-NEB near saddle points for flexible molecular systems.
• To improve the efficiency and robustness of transition state searches.

Main Methods:

• Developed an energy-weighted CI-NEB method by increasing spring constants with energy.
• Assessed performance on 121 molecular reactions.
• Combined CI-NEB with an eigenvector following method (NEB-TS) for accelerated convergence.

Main Results:

• Energy-weighted springs significantly improved saddle point location accuracy and efficiency.
• The NEB-TS method reduced average energy/force evaluations to 305.
• Achieved successful saddle point location in under 1000 evaluations on average.

Conclusions:

• The energy-weighted CI-NEB method offers superior resolution for flexible systems.
• The NEB-TS approach provides a robust and highly efficient strategy for transition state searches.
• An implementation is available in the ORCA software.