Probing Atomic Distributions in Mono- and Bimetallic Nanoparticles by Supervised Machine Learning

Abstract

Properties of mono- and bimetallic metal nanoparticles (NPs) may depend strongly on their compositional, structural (or geometrical) attributes, and their atomic dynamics, all of which can be efficiently described by a partial radial distribution function (PRDF) of metal atoms. For NPs that are several nanometers in size, finite size effects may play a role in determining crystalline order, interatomic distances, and particle shape. Bimetallic NPs may also have different compositional distributions than bulk materials. These factors all render the determination of PRDFs challenging. Here extended X-ray absorption fine structure (EXAFS) spectroscopy, molecular dynamics simulations, and supervised machine learning (artificial neural-network) method are combined to extract PRDFs directly from experimental data. By applying this method to several systems of Pt and PdAu NPs, we demonstrate the finite size effects on the nearest neighbor distributions, bond dynamics, and alloying motifs in mono- and bimetallic particles and establish the generality of this approach.

Formation and Functioning of Bimetallic Nanocatalysts: The Power of X‐ray Probes

“…[79,80] As the spatial resolution of state-of-theart X-ray nanoscopes (ca. [30,68] Methods which increase the sensitivity of at echnique,s uch as wavelet analysis [50] or MES, [49] can be used to efficiently extract apparently hidden information. [30,68] Methods which increase the sensitivity of at echnique,s uch as wavelet analysis [50] or MES, [49] can be used to efficiently extract apparently hidden information.…”

“…15 nm) still exceeds the dimensions of catalytically relevant NPs,m achine learning methods for obtaining the 3D NP structure show disruptive potential to revolutionize this field of research. [30,68] Methods which increase the sensitivity of at echnique,s uch as wavelet analysis [50] or MES, [49] can be used to efficiently extract apparently hidden information.…”

“…Under 100 mTorr reducing H 2 or oxidizing NO gas,t he NP surface composition changes,r esulting in increased Rh segregation (Figure 7b). Timoshenko et al [68] recently developed apowerful approach to reconstruct 3D atomic models of mono-and bimetallic NPs by combining WT XAS,supervised machine learning,a nd molecular dynamics simulations (Figure 7c). Deriving the 3D NP structure in real-time from in situ or operando XAS could revolutionize nanocatalysis.…”

“…c) neural network approach to derive the 3D NP structure based on WT EXAFS. Adapted from Refs [67,68]…”

Artificial Intelligence to Power the Future of Materials Science and Engineering

“…In recent years, AI has been applied in more and more fields, and ML research in the field of materials is rapidly developing, especially in that it can synthesize new materials and predict various chemical synthesis . In this section, we will explore how ML can help people solve the barriers between designing, synthesizing, and processing materials …”