Machine learning of atomic dynamics and statistical surface identities in gold nanoparticles

Area of Science:

• Materials Science
• Computational Chemistry
• Nanotechnology

Background:

• Metal nanoparticles (NPs) exhibit dynamic atomic behavior, influencing their properties.
• Characterizing atomic dynamics and atomic environments (AEs) in NPs is crucial but challenging.
• Understanding NP stability and atom mobility is key for realistic applications.

Purpose of the Study:

• To develop a machine learning approach for decoding intricate atomic dynamics in metal NPs.
• To identify and track the stability, survival, and interconversion rates of atomic environments (AEs).
• To provide a comprehensive picture of intrinsic atomic dynamics shaping NP properties.

Main Methods:

• Utilized machine learning to analyze high-dimensional data from molecular dynamics (MD) simulations.
• Developed an AEs' dictionary to label and track individual atoms within gold NPs.
• Monitored the emergence, annihilation, lifetime, and interconversion of AEs at various temperatures.

Main Results:

• Successfully decoded complex atomic dynamics in gold NPs using MD simulations and ML.
• Identified and quantified the stability and interconversion rates of native and non-native AEs.
• Established a method to estimate a "statistical equivalent identity" for metal NPs.

Conclusions:

• The machine learning approach effectively characterizes atomic dynamics in metal NPs.
• Tracking AEs provides crucial insights into NP stability and behavior.
• This method offers a comprehensive understanding of intrinsic atomic dynamics influencing NP properties.