Exploring the Importance of Dynamics in Metals Using Advanced Computational Methods

Understanding the dynamic behavior of materials at the microscopic level is essential for explaining their macroscopic properties. However, capturing complex atomic/molecular dynamics and translating them into a comprehensive understanding of material properties is challenging. This study focuses on metals, exploring the non-trivial atomic dynamics within them.

Initially, we investigate copper surfaces using deep-potential molecular dynamics simulations with a neural network potential trained on DFT calculations, resolving their complex dynamic behaviors. Additionally, we explore gold nanoparticles (NPs) to characterize their atomic dynamics at different temperatures. By extending our previous approach, we demonstrate that our analytical framework is applicable to various metal systems, providing insights into the intrinsic atomic dynamics shaping NP properties. This progress led to a collaboration with experimental groups, combining advanced computational and experimental techniques to characterize these systems with unprecedented resolution, overcoming the limitations of each method when used separately.

At the core of our approach, we utilize high-dimensional structural descriptors, such as SOAP, and unsupervised machine learning techniques to identify and track atomic environments across metal systems. This reveals the dynamic equilibrium between native and non-native AEs, offering a new perspective on the "statistical identity" of metals.

Overall, this general approach unravels and characterizes the intricate dynamics of metal systems. By examining microscopic dynamics, we can accurately determine macroscopic properties, enhancing our understanding of metal surface behaviors under relevant conditions. Shifting from static to dynamic perspectives, our approach advances the understanding of complex systems, providing comprehensive insights into the behavior of metals and other materials.