Graduation Year

2023

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Chemical, Biological and Materials Engineering

Major Professor

Clifford Henderson, Ph.D.

Co-Major Professor

John J. Kuhn, Ph.D.

Committee Member

Peter P. Ludovice, Ph.D.

Committee Member

Lawrence L. Stern, Ph.D.

Committee Member

Nathan N. Gallant, Ph.D.

Keywords

Force-field Parametrization, Coarse-graining, Atomistic Simulation, Transferability, Glass Transition Temperature

Abstract

Polymers are in a wide range of material forms with various resources, including natural to synthetic processes. Polymers consist of covalently bonded monomers to each other. The chain length of polymers varies from a few monomers to thousands of monomers. This wide range of chain length enables polymers with attractive capabilities of chemical and physical properties. Thus, polymers are highly attractive in different scientific and industrial applications such as drug delivery, optical coatings, catalysts, semiconductors and organic photovoltaics. This work has combined updated accurate methods for accurate initial conformations, and modified force field parameters to produce a generalized method for accurate simulation of amorphous polymers. This method was successfully applied to the simulation of atactic polystyrene and polypropylene. Also, the combination of mapping from an accurate atomistic simulation, and modification of coarse-grained (CG) force-field parameters, produced the first CG mapping procedure that can accurately reproduce the glass transition as well as properties both above and below the glass transition.

The macroscopic behavior of the polymers strongly depends on the length of polymers. Polymers are long chains and because of their chain length and entanglement, they are high density and viscous materials. Thus, when a deformation is applied on them, the polymers response to the effect of the deformation strongly depends on the rate of the deformation. This indicates that polymer chains properties are included in a broad range of time and size scales compared to small molecule materials. In order to explore polymer characteristics, we require to learn the details of polymer chains behavior and the consequences of chain entanglements that produce some limitations for chain movements. Computer simulations at atomistic level could educate us about the details of molecular behavior of polymer chains in the glassy and melt states.

In order to explore and evaluate mechanical and thermodynamic properties comparable to real polymers behavior, there is a strong need to develop simulations for a long time and length scales to relax and equilibrate polymer systems. Thus, the atomistic simulations are computationally time consuming and expensive to solve the equations of motion for a system consisting of thousands of particles that interact with each other at the atomistic level. One solution to overcome this major restriction of atomistic simulations is mapping the atomistic structures of polymer chains to the CG level of them. In the CG level, a group of atoms are mapped to a single group, which is called a super-atom or a bead. Mapping the atomistic structures to the CG level reduces the total degrees of freedom and makes the system simpler in comparison to atomistic simulations, which significantly accelerates the computational process compared to detailed atomistic or united atom simulations for polymer materials. As a consequence, the running time and length scale of simulation capability could be multiplied by order of magnitude.

However, there was a strong need for CG models that are capable to capture the material properties that connected to the details of the monomer structures. It means that even defining the CG system improves the efficiency of the simulation in terms of simulation speed and simulation size, still we need to define an accurate CG model that can present the nature of the atomistic system that was mapped to CG level. Although CG models are faster than atomistic simulations, there are still several gaps between the atomistic and CG models for polymer materials in terms of time and length scale that must be addressed. Therefore, CG models experience several challenges and limitations like representability and transferability. Representability means the ability of CG models to predict desired properties other than the properties that were used to develop that CG model at the specific thermodynamic state that the model was developed. On the other hand, transferability explains the CG models capability to predict desired properties at different thermodynamic states other than the state that the properties were developed. For instance, transferability evaluates the CG ability to predict the experimental properties of polymer for different chain lengths and thermodynamic states like various temperatures and pressures.

Although many coarse-grained models have been developed to describe polymers behaviors, still most of them are: (1) not generated from accurate atomistic models, and (2) only able to qualitatively predict trends and behavior of such systems. In this work, coarse-grained polymer models were developed to improve the capability of CG model to predict experimental properties related to structure or thermodynamic characteristics of the polymer materials to overcome some of these limitations and make more quantitatively accurate predictions of polymers behavior possible. Such models will be developed using a combination of models using both experimental data and atomistically detailed polymer simulations.

One challenge that must be overcome in these efforts is that equilibrating amorphous polymers into realistic conformations is itself a challenge since in their dense and highly viscous state the polymer chains have very low diffusivities. Therefore, it takes a very long time to relax and equilibrate the polymer chains, specifically for a large number of particles in the simulation box. One solution to overcome these drawbacks is developing methods to efficiently produce an accurate initial conformation, which are explored in this work. The resulting atomistic simulations will be coupled with experimental data to train and develop coarse-grained polymers that can capture the behavior of such systems in an accurate manner.

The objective of the proposed work is to address these challenges by exploring the following questions for atomistic and coarse-grained simulation:1. Does the initial generator method control the structures, dynamics, physical properties, and simulation time for polymeric systems? If yes, what is the most efficient method to build initial conformations? 2. What is the force-field parameters set that can predict the physical and thermodynamic properties of polymers? 3. What level of coarse-graining can represent realistic and atomistic behavior for polymers? 4. How well is transferable and representable the coarse-grained model for polymer chains to predict the structures, dynamics, and physical properties accurate in a wide range of temperature?

This dissertation is structured as follows. Chapter 1 provides an overview of molecular dynamics simulations for both atomistic and coarse-grained simulations for desired polymers in this study. In Chapter 2, the methodology for generating initial guess conformations of polymer chains at the atomistic level is discussed in detail. The primary focus is the development of a methodology for constructing realistic atomistic structures of polymer chains and optimizing force-field parameters for two specific polymers: atactic polystyrene (aPS) and atactic polypropylene (aPP) which is detailed in Chapter 3. In Chapter 4, the CG models and mapping methods details for the aforementioned polymers are presented. CG models developed to simulate polymer chains of aPS and aPP and the structural and thermodynamic properties are compared with atomistic simulations and experimental values. Chapter 5 provides conclusions and recommendations for future research works.

Download

Included in

Engineering Commons

Share

COinS