Diffusion Growth Mechanism of Nanowires with Multi-scale Theory and OptiBoost Method for Accelerated Molecular Dynamics

Open Access
- Author:
- Cui, Jianming
- Graduate Program:
- Chemical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 29, 2024
- Committee Members:
- Kristen Fichthorn, Chair & Dissertation Advisor
Michael Janik, Major Field Member
Adri van Duin, Outside Unit & Field Member
Robert Rioux, Professor in Charge/Director of Graduate Studies
Rui Shi, Major Field Member - Keywords:
- Nanowires
Molecular Dynamics
Multi-scale Theory
Markov Chains
Accelerated Molecular Dynamics
Hyperdynamics
Bond-boost method
OptiBoost
Twinning
Detwinning - Abstract:
- Crystals with penta-twinned structures can be produced from diverse fcc metals. Various penta-twinned nanocrystal shapes can be synthesized, including nanowires (NWs) nanorods (NRs) and various types of decahedra (Dh). However, the mechanisms that control the final product shapes are still not well understood. Several approaches were made to explain and predict the final shapes of the products and the mechanism of the morphological evolution. The theory of absorbing Markov chains was used to account for the growth of penta-twinned decahedral seeds via atom deposition and surface diffusion. In the case of Ag seeds with uncapped surface, we predicted the formation of various types of products: decahedra, nanorods, and nanowires. We showed that the type of product depends on the morphology of the seed and that small differences between various seed morphologies can lead to significantly different products. In the case of chloride- and alkylamine-mediated, solution-phase growth of penta-twinned Cu nanowires, we calculate coarse-grained, interfacet rates for nanowires of various lengths up to hundreds of micrometers and diameters in the 10 nm range. We predict nanowires with aspect ratios of ∼100, based on surface diffusion alone. We also account for the influence of a self-assembled alkylamine layer that covers most of the {100} facets, but is absent or thin and disordered on the {111} facets and in an “end zone” near the {100}/{111} boundary. With an end zone, we predict a wide range of nanowire aspect ratios in the experimental ranges. The theory of Markov Chains is effective to coarse-grain the results from molecular dynamics (MD) and density-functional theory (DFT) to the meso-scale, which makes it possible to compare the simulations results with experimental results. However, the whole configuration of the system is required for the Markov matrix with transient states, absorbing states, and transitions between states. In the situation of complex systems, accelerated Molecular Dynamics (aMD) is more feasible to improve the efficiency of traditional MD simulations. AMD simulations based on hyperdynamics (HD) can significantly improve the efficiency of MD simulations of condensed-phase systems that evolve via rare events. However, such simulations are not generally easy to apply since appropriate boosts are usually unknown. In this work, we developed a method called OptiBoost to adjust the value of the boost in HD simulations based on the bond-boost method. We demonstrated the OptiBoost method in simulations on a cosine potential and applied it in three different systems involving Ag diffusion on Ag(100) in vacuum and in ethylene glycol solvent. In all cases, OptiBoost was able to predict safe and effective values of the boost, indicating that the OptiBoost protocol is an effective way to advance the applicability of HD simulations. The morphological evolution of seeds on which the final product shape depends is critical, such as twinning and detwinning of the penta-twinned structure. MD simulations were performed to investigate details of the detwinning process, which confirm the experimental results and reveal how strain and surface migration play a role in either the twinning or detwinning of these nanoparticles. The detwinning of asymmetrical penta-twinned Au nanoparticles is complicated and involves various dislocation activities on twin-boundaries (TBs), along with atomic surface diffusion. We find that TBs likely migrate in groups (i.e., one layer of plane gliding triggers another) due to dislocation reactions among various TBs. TBs migrate back and forth during detwinning, especially when the 5-fold axis is far away (≳3 layers) from the periphery of a 5-FT nanoparticle due to the competition between relaxing strain energy and decreasing surface energy: i.e., detwinning relaxes strain energy but increases surface energy (more {100} surface area), while twinning decreases surface energy (more {111} surface area) but increases strain energy. Each atomic layer migration appears to possess a corresponding energy barrier. The asymmetrical lattice tensile strain distribution, especially tensile strain that is perpendicular to the TBs, is the dominating driving force for dislocation slipping, TB plane gliding, and thus detwinning.