A Computational Framework Enabling Fast Simulation and Predictive Pathway Control of Self-assembling Particulate Systems
Abstract
The objective of this project is to establish a new computational framework enabling fast while accurate predictions for self-assembling particulate systems, and by integrating with experiments, to design and realize new photonic materials from nanoparticles (NPs). The new computational framework will harvest the power of advanced machine learning techniques and be built on graph neural networks (GNNs). By surrogate modeling through GNN, significantly faster simulations with desired accuracy will be enabled for predicting the dynamics of particles subject to external drive, interparticle potentials, hydrodynamic interaction, and thermal fluctuations. Besides efficiency and accuracy, the following attributes are featured for the proposed GNN-based computational framework: requiring low training cost, transferable to suspensions of any numbers of particles and across different directing forces, applicable to anisotropic-shaped particles, and scalable to large systems. These advanced features will lead the proposed computational framework into a powerful predictive tool for fundamental studies and computer-aided design of self-assembling particulate systems. In addition to theoretical verification, systematic experimental validations will be performed using both micron-sized colloids and NPs of 30 Ð 100 nm. The comprehensively validated computational framework will finally be integrated with the liquid-phase transmission electron microscopy experiments to study the assembly of anisotropic NPs and to achieve predictive pathway control for NP assemblies with exotic photonic properties. As a demonstration, we will target at two novel gold NP-based assembled structures of chiral meta-lattices. Their chiroptical properties are essential for applications in chiral discrimination of molecules, polarization sensitive imaging, machine vision, and enantiomer-selective catalysis.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 28, 2023
- Source ID
- W911NF2310256
Entities
People
- Wenxiao Pan
Organizations
- Army Contracting Command
- United States Army
- University of Wisconsin–Madison