An Efficient Growth Pattern Algorithm (GrowPAL) for Cluster Structure Prediction.

Area of Science:

• Computational Chemistry
• Materials Science
• Nanotechnology

Background:

• Identifying the lowest energy isomers in large atomic clusters is computationally challenging.
• Existing methods often require extensive optimization, increasing computational costs.
• Understanding cluster growth pathways is crucial for designing novel materials.

Purpose of the Study:

• To introduce a novel algorithm, the Growth Pattern Algorithm (GrowPAL), for efficient global minima identification in atomic clusters.
• To evaluate GrowPAL's effectiveness on various cluster systems, including Lennard-Jones, Sutton-Chen, and boron clusters.
• To analyze the algorithm's performance and identify potential growth pathways.

Main Methods:

• GrowPAL generates initial seeds by adding an atom to a smaller cluster via an interstitial-type (I-type) addition mechanism.
• The algorithm was tested on Lennard-Jones (LJ) clusters up to 80 atoms, including challenging minima like LJ38 and LJ69.
• A deconstruction scheme was used to analyze GrowPAL's advantages, limitations, and identify 'forebear' structures for studying growth.

Main Results:

• GrowPAL successfully identified challenging global minima in LJ clusters with fewer optimizations than existing methods.
• Application to Sutton-Chen clusters (5-80 atoms) revealed three new lowest energy forms.
• GrowPAL accurately identified all reported minima for boron clusters (8-15 atoms).

Conclusions:

• GrowPAL provides a practical and efficient solution for identifying global minima in hierarchical atomic systems.
• The algorithm significantly reduces computational costs associated with cluster structure prediction.
• GrowPAL facilitates the exploration of complex cluster landscapes and the discovery of new stable structures.