Gold Segregation Improves Electrocatalytic Activity of Icosahedron Au@Pt Nanocluster: Insights from Machine Learning^†

Abstract

Main observation and conclusion As a common electrocatalytic system, Au‐Pt alloy particles are often prepared as Au‐core‐Pt‐shell (Au@Pt) to make full use of platinum. However, Au has a strong tendency to segregate to the outer surface, leading to the redistribution of the active sites. Unfortunately, the mechanism of such reconstruction and its effect on the electrocatalytic activity have not been thoroughly discussed, largely owing to the complexity of in‐situ characterization and computational modeling. Herein, by taking the 55‐atom Au13Pt42 core‐shell nanocluster as an example, we utilized the neural network potential at density functional theory (DFT) level and the genetic algorithm to search the complex global configurational space. It turns out that it is thermodynamically favorable when all gold atoms are segregated to the surface and the shape of the cluster tends to change from icosahedron to a distorted amorphous structure (at a reduced core, DRC) with a unique gold distribution. Towards understanding the dynamic activity variation of oxygen reduction reaction (ORR) on this bimetallic Au@Pt system, oxygen adsorption energy calculations show that this reconstruction could not only increase the number of adsorption sites but also dramatically improve the ORR catalytic activity of each site, thus enhance the overall ORR reactivity.

Achieving Theory–Experiment Parity for Activity and Selectivity in Heterogeneous Catalysis Using Microkinetic Modeling

“…66 Chen et al applied the neural network potential and the genetic algorithm to identify the most stable configuration of Au@Pt, which is a common electrocatalytic system. 67 Li et al studied the Pd−Ag−H system for the acetylene hydrogenation reactions using the global neural network potential. 68 Constructing an accurate PES can be time-consuming and resource-intensive, but the rapid development of machine learning has provided a possible pathway for building microkinetic models on more complex or amorphous surfaces.…”

“…Xu et al proposed an on-the-fly machine learning method to accelerate AIMD simulations for adsorption energy estimations . Chen et al applied the neural network potential and the genetic algorithm to identify the most stable configuration of Au@Pt, which is a common electrocatalytic system . Li et al studied the Pd–Ag–H system for the acetylene hydrogenation reactions using the global neural network potential .…”

Kernels‐Different Au25Nanoclusters Enhanced Catalytic Performance via Modification of Ligand and Electronic Effects

“…The continuous density of states breaks up into discrete energy levels, which translates into the unique optical, electrical, and chemical properties in comparison with nanoparticles. [ 6‐11 ] Increase of unsaturated sites on the surface (surface effect) enables the clusters to exhibit good catalytic activity and unique selectivity in reactions such as oxidation ( Au 38 , Au 144 ), hydrogenation ( Cu 25 ), C—C coupling ( Au 11 , Cu 53 ), A 3 coupling ( Au 13 ), cycloaddition ( Cu 20 ), etc . [ 12‐17 ] Nanoclusters with precise structures can also serve as model catalysts to reveal the correlation between catalyst performance and structure.…”

Structural characterisation of nanoalloys for (photo)catalytic applications with the Sapphire library

“…properties of the MNP itself 1,[15][16][17] Focusing on catalytic applications, the role of the MNP's and NA's surface is central. In this context, electronic structure calculations represent an established route to infer the structure-property relationships which rule the activity, selectivity, and stability of the catalyst [18][19][20][21] . A knowledge of robust structure-property relationships, and of the finite temperature probability of observing MNPs and NAs with a given architecture, in turn, allows to draw design rules, to predict the activity, and to forecast the ageing of a nanocatalysts.…”