Datenpaket: Dataset for Publication "Elucidating Mn Promoter Structures and Stability on Co Nanoparticles through Machine Learning Potential-powered Genetic Algorithm"

Dataset including genetic algorithm scripts, fine-tuned CHGNet in .tar format, global minima structures in .xyz format, plots (.png) showing algorithm progression (Energy vs. # of candidates), database files in .db format containing all relaxed candidates for all genetic algorithm runs, training dataset both in .db and .json format and structural files of the MnxOyHz motifs on different fcc-Co surface conformations extracted from global minima structures

The folders 6_nm_fcc, 8_nm_fcc, and 8_nm_hcp contain subdirectories corresponding to the investigated Mn phases. Each subdirectory includes plots illustrating the genetic algorithm convergence (e.g., energy vs. candidate) and the structural files (.xyz) of the identified global minima.
The "trajs" folder contains the databases of the trajectories of all candidates screened during the genetic algorithm searches for each run.
The "datasets" folder contains the training and test datasets used for the fine-tuning and benchmarking of the machine-learning potential on both fcc and hcp surfaces. The datasets are provided in both .db and .json formats. This folder also includes the fine-tuned machine-learning potential in .tar format.
The "ga_scripts" folder contains the Python scripts based on the ASE genetic algorithm (ase.ga) framework that were modified and extended specifically for the present study.
Finally, the "Mn_structural_motifs" folder contains the structural files and optimization trajectories of representative Mn structural motifs adsorbed on different Co surface sites, extracted from the identified global minimum structures.