Optimized energy landscape exploration using the ab initio based activation-relaxation technique.

Area of Science:

• Computational Materials Science
• Theoretical Chemistry
• Condensed Matter Physics

Background:

• Unbiased open-ended methods are crucial for understanding diffusion and relaxation mechanisms.
• Previous methods demanded significant computational resources, limiting their use with ab initio packages.
• Transition state searches are vital for defect diffusion, growth processes, and catalysis.

Purpose of the Study:

• To revisit and enhance the Activation-Relaxation Technique (ART nouveau) for efficient transition state searches.
• To couple ART nouveau with the BigDFT electronic structure code for ab initio calculations.
• To systematically study potential energy surfaces and reaction pathways in materials.

Main Methods:

• Introduced a two-step convergence to the saddle point, combining Lanczós algorithm with direct inversion in interactive subspace.
• Coupled the enhanced ART nouveau with BigDFT, a parallelized Kohn-Sham density functional theory code using wavelet basis sets.
• Achieved an average of 300-700 force evaluations per successful event for generating reaction pathways.

Main Results:

• Successfully generated transition states and reaction pathways with significantly reduced computational cost.
• Applied the method to study C(20) clusters, vacancy diffusion in silicon, and 4H-SiC surface reconstruction.
• Demonstrated the efficiency and applicability of ART nouveau coupled with BigDFT for complex materials systems.

Conclusions:

• The enhanced ART nouveau provides an efficient and systematic approach for unbiased transition state searches.
• This method enables the study of complex phenomena like defect diffusion and surface reconstruction using ab initio calculations.
• The integration with BigDFT offers a powerful tool for exploring materials' potential energy surfaces.