Localized saddle-point search and application to temperature-accelerated dynamics.

Area of Science:

• Computational materials science
• Surface science
• Chemical physics

Background:

• Temperature-accelerated dynamics (TAD) simulations are crucial for studying activated processes in materials.
• Accurate determination of energy barriers is computationally intensive for large systems.
• Existing methods often face challenges with system size scalability.

Purpose of the Study:

• To introduce a novel method for accelerating TAD simulations.
• To reduce the computational cost of saddle-point searches in large systems.
• To analyze the accuracy and performance of the proposed method.

Main Methods:

• Implementation of a localized saddle-point (LSAD) search within TAD simulations.
• Localization of energy barrier calculations to a small subset of atoms involved in a transition.
• Validation using model systems including Ag(100) and Cu(100) surfaces, Cu radiation damage, and metal heteroepitaxial growth.

Main Results:

• Achieved N-independent scaling for saddle-point search computational cost with system size N.
• Demonstrated significantly improved performance for Ag/Ag(100) annealing and Cu/Cu(100) growth simulations.
• Quantified and confirmed negligibly small errors in energy barrier calculations due to localization.

Conclusions:

• The LSAD method offers a substantial performance enhancement for TAD simulations.
• This approach effectively addresses the scalability limitations of traditional methods.
• LSAD provides a computationally efficient and accurate tool for materials simulation.