Stable structure optimization of Au-Cu-Pt trimetallic nanoparticles based on genetic algorithm

Abstract

Alloy nanoparticles exhibit multifunctional properties different from monometallic nanoparticles. Especially, when a third metal is introduced into bimetallic nanoparticles system to form trimetallic nanoparticles, their chemical activities will be further improved. As the catalytic reaction of nanoparticles usually takes place on surfaces, and the activity and stability are closely related to their structures, therefore the research on the stable structure is crucial for understanding their catalytic activities. In addition, the electrochemically synthesized tetrahexahedral nanoparticles bound with highindex facets may exhibit greatly enhanced catalytic activity because of their large density of low coordination sites at the surface. Based on the above reasons, this paper carries out the investigation on the stable structures of tetrahexahedral Au-Cu-Pt trimetallic nanoparticles by using an improved genetic algorithm and the quantum-corrected Sutton-Chen (Q-SC) type many-body potentials. To avoid the genetic algorithm being trapped into premature convergence, two improvement strategies are developed. On the one hand, an atom coordinate ranking operation, which is implemented according to the atomic distance from the core, is proposed for reducing the probability of individual loss. On the other hand, an alternating bit means is introduced into the crossover operation to keep the atomic composition ratio unchanged. Moreover, the performance of genetic algorithm and the influence of original configuration on the stable structures of Au- Cu-Pt trimetallic nanoparticles with different sizes and different compositions also have been investigated. One stochastic distribution structure and three core-shell distribution structures of Au@CuPt, Cu@AuPt and Pt@AuCu are adopted as the initial structures, respectively. Eleven optimization trials on Au-Cu-Pt trimetallic nanoparticles in Au-Cu-Pt system with Au : Cu : Pt of 0:343 : 0:343 : 0:314 with 443 atoms are used to verify that the different original structures should have no effect on the final stable structure. Furthermore, 30 random trails on Au-Cu-Pt trimetallic nanoparticles at Au : Cu : Pt of 0:316 : 0:316 : 0:368 with 443 atoms are conducted to prove that the genetic algorithm can obtain robust results with small standard deviation. Finally, the segregation analysis results show that: In Au-Cu-Pt trimetallic nanoparticles, Au and Cu atoms prefer to aggregate on the surface while Pt atoms are preferential to locate in the core. Furthermore, Cu atoms exhibit stronger surface segregation than Au atoms. For small Au or Cu concentration, Au and Cu atoms would display the maximum segregation. They begin to compete during aggregation, and the Cu atoms have a strong tendency for surface segregation when the number of Au and Cu atoms is bigger than the total number of surface atoms. With increasing number of Au and Cu atoms over those on the surface and sub-surface, Au atoms would display a strong surface segregation than Cu atoms. Additionally, Cu atoms will mix with Pt atoms in the inner layers over the sub-surface after occupying the surface. The distribution of surface atoms has been further examined by the analyses of coordination number: the Cu atoms tend to occupy the vertices, edges and kinks, while the Au atoms preferentially segregate to the flattened surface. This study provides a perspective on structural features and segregation behavior of trimetallic nanoparticles.