Searching for low-energy structures of nanoparticles: a comparison of different methods and algorithms

TLDR

Nanoparticles can exhibit low‑symmetry or non‑crystalline shapes, and their structure governs physical and chemical properties, making the global optimization of their high‑dimensional potential energy surfaces a challenging computational problem. The study depicts the landscape of global optimization strategies for nanoclusters, emphasizing genetic algorithms and Basin‑hopping methods. The authors compare and enhance Basin‑hopping efficiency by optimizing test systems of varying size and composition.

Abstract

Nanoparticles can have unusual, low symmetry or non-crystalline shapes. Since structure determines nanoparticle physical and chemical properties, many efforts have been devoted to predict what are the most stable structural motifs depending on cluster size and composition. The global optimization of the 3N-dimensional potential energy surface of a nanocluster is nevertheless a very difficult computational problem. Here we depict the scenery of the global optimization strategies applied to the study of nanoclusters, focusing on genetic and Basin-hopping approaches. Moreover, several strategies to improve Basin-hopping efficiency are discussed and compared through the optimization of test-systems with different size and composition.