Anisotropic Assembly of Nanoparticles Explored through Molecular Dynamics Simulations, Global Optimizations and Machine Learning Methods

Novel applications of polymer nanocomposites like plasmonics often require anisotropic organization of nanoparticles (NPs) in polymers. However, achieving unique anisotropic assemblies of NPs in polymers is challenging since NPs tend to self-assemble into three-dimensional close-packed aggregates. In this work, we tackle this challenge of achieving anisotropic NP assembly in polymers through a combination of molecular dynamics (MD) simulations, global optimization, and machine learning. First, we present a new strategy for assembling NPs into anisotropic architectures in polymer matrices, which leverages the interfacial tension between two mutually immiscible polymers forming a bilayer and differences in the relative miscibility of polymer grafts with the two polymer layers to confine NPs within 2d planes parallel to the interface.¹ Through coarse-grained (CG) MD simulations, we demonstrate this strategy, showcasing the assembly of NP clusters, such as trimers with tunable bending angle and anisotropic macroscopic phases, including serpentine and branched structures, ridged hexagonal monolayers, and square-ordered bilayers. The above MD simulations are however inefficient for determining the equilibrium structures of NP assemblies, particularly those involving many particles or complex unit cells. To address this issue, we adapt the efficient Basin-hopping Monte Carlo algorithm to locate the global minimum energy configurations of NPs at interface, allowing us to explore the full breadth of NP structures possible at interface and uncovering many unique NP structures.² While exploring the assembly of polymer-grafted NPs at polymer interfaces using explicit CG MD simulations, we observe that many-body effects play an important role in the formation of quasi-1d structures. However, explicit modeling of the polymer grafts and melt chains is highly computationally expensive, even using CG models. We thus introduce a machine learning approach to develop an analytical potential that can describe many-body interactions between polymer-grafted NPs in a polymer matrix. The developed potential reduces the computational cost by several orders of magnitude and thus allows us to explore NP assembly at large length and time scales. Overall, the anisotropic NP structures discovered in this study hold significant potential for applications in various fields, including plasmonics, electronics, optics, and catalysis, where precise and anisotropic NP arrangements within polymers are essential for achieving desired functionalities.<br/><br/>[1] Tang, T. Y., Zhou, Y., & Arya, G. (2019). Interfacial assembly of tunable anisotropic nanoparticle architectures. ACS nano, 13(4), 4111-4123.<br/>[2] Zhou, Y., & Arya, G. (2022). Discovery of two-dimensional binary nanoparticle superlattices using global Monte Carlo optimization. Nature Communications, 13(1), 7976.<br/><br/>Portion of the work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract Number DE-AC52-07NA27344.