Search

This dissertation outlines research quantifying the entering and exiting the quinone reaction chamber in NADH dehydrogenase or respiratory complex I. Respiratory complex I, the first complex in the respiratory electron transport chain. The respiratory electron transport is essential for all aerobic life. The methods used to quantify the entrance and exit process are geometric modeling, steered molecular dynamics, and singular value decomposition of the process. Five structures were analyzed: bacterial, yeast, ovine mt, mice mt, and human complex I. The structures reveal an almost 30 angstrom tunnel-like chamber for quinone binding in the core part of the enzyme, at the joint between the membrane and hydrophilic arms of the enzyme. The entrance of this quinone chamber located in ND1 subunit and has an apparent bottleneck of quinone/quinol passage. The first chapter introduces complex I and how transition state theory using diffusion kinetics gives an approximate maximum for the energy of crossing the bottleneck for quinone/quinol passage. Chapter 2 introduces the techniques used to quantify the difficulty of passage as well as methods to identify modes for collective confirmational changes for bottleneck opening. Chapters 3 and 4 are reproductions of the published papers based on this work. The appendices are reproductions of the supplemental information for those two papers.

Dynamic Mechanical Analysis (DMA) is an important experimental characterization technique for the mechanical properties of polymers. In this paper, we translate such a technique into a Molecular Dynamics (MD) simulation workflow capable of outputting properties from DMA within a much wider frequency and temperature range than what can be experimentally observed. We also introduce metrics for convergence of simulation runtime and DMA straining size that are not discussed in comparable computational literature but that drastically affect the accuracy of results. We validate the efficacy of the work by comparing our obtained storage and loss modulus data on Polyvinylidene Fluoride (PVDF) with previous computational literature and indicating performance and accuracy improvements for such study based on our convergence metrics. Finally, we perform a complete workflow run at GHz-range frequencies on PVDF utilizing the All-Atom Optimized Potential for Liquid Simulations (OPLS-AA). With this, we hope to validate and justify this forcefield as a good standard for high-throughput analyses on other polymers with the workflow and to compare our data with experimental results employing the Time-Temperature Superposition (TTS) method.

There were three main driving forces behind this thesis: global concern over climate change mainly due to uncontrolled carbon dioxide (CO2) emissions, the excitement over the discovery of graphene and its versatile potential, and the potential to design three-dimensional (3D) or two-dimensional (2D) structures, in our case using unique triptycene molecule. We examined two polymeric materials for CO2 adsorption and suggested simple design of disordered carbons suitable for gas adsorption studies. The approach in each task was to examine structural and adsorption properties of materials using detailed atomistic modelling employing Monte Carlo and Molecular Dynamics techniques and where possible provide experimental measurements to validate the simulations. The thesis is presented as a collection of papers and the work can be divided into three independent projects. The aim of the first project is to utilize graphene as an additive in polymer composites in order to increase separation between the polymer chains increasing available surface area. The matrix used is a polymer of intrinsic microporosity (PIM-1), which possess large surface area and narrow nano-sized ( > 2nm) pore distribution attractive for gas separation membrane applications. Adding a filler can reduce aging of the polymer, and enhance permeability across the membrane, often to the expense of loosing selectivity. Therefore, we investigated the packing of PIM-1 chains in presence of discrete 2D graphene platelets and 3D graphene-derived structures and its effect on composite structure and adsorption properties. We found that additives do not alter structural polymer properties at the molecular level preserving the same adsorption capacity and affinity. Potential permeability increase would benefit from the retention of selectivity in the material. Building on design philosophy of materials with intrinsic microporosity we continued further investigation of 3D graphene-derived structures. The idea is that highly concave molecules or polymer chains pack inefficiently creating microporous materials with sufficient surface area for gas adsorption. 3D propeller-like structures were derived from graphene arms connected through the rigid triptycene and other types of cores. The resulting structures created a large amount of micropores and showed similar CO2/CH4 selectivity to activated carbons reported in the literature. It was shown that rigid triptycene core leads to more open structures. The model was also applied to model commercially available activated carbon to predict n- perfluorohexane adsorption. The fitting to experimental structural information proved to be challenging due to trial and error nature of the approach. Nevertheless, the simple packing procedure and diverse structure design have a great potential to serve as a virtual model for porous carbons that possess pore complexity and does not require any previous experimental data to be build on. The last project concerns CO2 adsorption and selectivity over CH4 and N2 in recently reported triptycene-based polymer. The triptycene shape polymer can form a porous 2D network that can be exfoliated into free-standing sheets and potentially used as a membrane. Sheets stack in the bulk material forming anisotropic channel pores. Additionally it contains fluoro- functional groups, which are known to have a high CO2 affinity. We explored pore structure and chemistry of stacked material for gas adsorption and predicted comparable capacity and CO2 selectivity to other microporous covalent materials such as activated carbons and PIMs. The CH4/N2 selectivity was similar to currently most selective material belonging to MOF family. We showed that fluoro-group have a positive effect on CO2 affinity, however predictions are sensitive to the charges of fluorine atoms assigned by different methods.

Molecular Modeling for 3D Printing and Biological Applications Abstract This dissertation covers several molecular modeling efforts applied across length and time scales. These techniques are primarily applied to two major categories: 3D printing (chapters 2-4) and biological applications (chapters 5-7). These computational efforts were part of a collaborative effort to improve the experimental design process for these two applications and have corresponding experimental validation. We first present the application of multi-scale simulation techniques to the process of controlled assembly, which is a critical pre-cursor to enabling 3D printing at the nano-scale. This controlled assembly process depends on the complex interplay between surface, solute, and solvent interactions during the evaporation process. In order to actually model the assembly that takes place during this evaporation process a computational software and methodology had to be developed to replicate evaporation of mixed solvent systems and account for potential hydrogen bonding shells. This resulting methodology was tested for different evaporation rates and scales and is presented in Chapter 2. Chapters 3 and 4 are application papers using this methodology. Chapter 3 focuses on the assembly of small linear sugar molecules analyzed using atomistic molecular dynamics simulations. This atomistic resolution allows the probing of phase transitions these sugars undergo as they assemble into various morphological structures during evaporation. Chapter 4 focuses on the coarse-grained interactions between the phospholipid POPC, various surface types, and mixed solvents. This work investigated the specific contributions of surface type and solvent type on assembled features as a function of evaporation. The final chapters of the work 5-7, focus on the virus that causes COVID-19, SARS-CoV-2. While no longer a part of an additive manufacturing process, these chapters remain a part of a collaborative effort to apply computational tools towards the experimental design process of a specific application, the development of a diagnostic or therapeutic tool. These chapters focus on the evaluation of SARS-CoV-2 spike protein, and specifically the receptor binding domain (RBD) and its interactions with the human angiotensin-converting enzyme 2 (hACE2) receptor. These structures were simulated using atomistic molecular dynamics, evaluated for the structural stability under different glycosylations that would derive from different synthesis methods for treatment (chapter 5). The interactions between the two during binding were evaluated using molecular force pulling simulations, with specific focus on analyzing the hydrogen bonding, electrostatic, and van der waals interactions between the two (chapter 6). This work demonstrated that they glycosylation contributes directly and indirectly to strengthening the interaction with hACE2. Finally the structural stability of various spike truncations synthesized in Chinese hamster ovary cells for use in therapeutic / diagnostic applications were analyzed in chapter 7.

Block copolymers self-assemble to form a variety of different phases with highly regular patterns, depending on the microscopic ordering of molecules. Paramount to understanding and controlling this "order" is to have good "order parameters", variables that can be used to track the changes occurring in the system as it transitions from disorder to order. Some common phases that block copolymers form include the lamellar, cylinder and gyroid network. In this paper, we use molecular dynamics to simulate the growth of these phases from an isotropic liquid. A binary nanoparticle mixture model is used that allows the simulation of systems with large enough sizes for the desired phase to nucleate and grow. We develop local order parameters based on particle symmetries and geometrical constraints that can identify and track the nucleation and growth of ordered domains along the transition pathway. We put forth a framework that could be extended towards understanding the self-assembly of other bicontinuous phases such as the double diamond or plumber’s nightmare, as well as prove useful in estimating free energy barriers and nucleation rates using rare-event sampling techniques.

An accurate prediction of the rate of dislocation motion is key to the fidelity of multi-scale plasticity models of metals and alloys. In this dissertation, atomistic simulations and rate theories based on statistical mechanics are used to accurately predict the rate of three main dislocation motion mechanisms: 1) Dislocation motion across precipitates 2) Dislocation motion through a field of obstacles 3) Dislocation motion via kink-pair nucleation For these mechanisms, the accuracy of both conventional and modern rate theories is examined by comparing their predictions to benchmarks obtained from MD simulations. Different variants of the Harmonic Transition State Theory, as the most common rate theory in the literature, are found to provide grossly inaccurate predictions for all three problems. It is shown that the inaccuracy of these approaches stems from their assumptions about the entropy barrier. The original version of HTST estimates the entropy barrier by the harmonic vibrational entropy, which is found to be inaccurate for all three problems due to thermal softening. Other versions of HTST based on simple estimates of the attempt frequency consider smaller values for the vibrational entropy, and hence provide even more inaccurate predictions. Furthermore, all variants of HTST neglect the configurational entropy, which turns out to be significant for the kink-pair nucleation problem. The utility of the Finite Temperature string method for computing a reaction coordinate and a free energy profile was examined for the three problems. The method provides an accurate reaction channel for dislocation-obstacle interactions but fails to provide a reasonable free energy profile. The reasons are investigated and discussed in the first paper presented in this dissertation. The original version of the method fails to provide a reaction channel for the kink-pair nucleation problem because it has not been designed for problems with multiple reaction channels. To address this issue, a modification to the approach based on physical intuitions about the problem is proposed and is shown to be effective. Different variants of the Transition Interface Sampling approach, as a modern rate theory, are found to be capable of accurately predicting the rate for all three problems. TIS and its Path Swapping version are found to be effective for dislocation-precipitate interactions. The method is also accurate in the jerky motion regime of dislocation motion through a field of solutes. For the smooth motion regime, however, the Partial Path version of TIS --- designed for diffusive barriers --- had to be used. In order to provide accurate predictions for the kink-pair nucleation problem, TIS had to be modified based on physical intuitions about the problem because the method has not been designed for problems with multiple reaction channels such as the glide of screw dislocations in BCC transition metals. The performance of the Meyer-Neldel (MN) rule, as the most common entropy estimation approach in material mechanics, is examined for the three problems. It is shown that the method accurately predicts the entropy barrier for dislocation-precipitate interactions but fails to fully explain the entropy barrier for the other two problems. The assumptions and theoretical justification of the method for dislocation processes, which are often neglected in the literature, are revisited and simple alternative models are proposed.

Graphene is a single layer two-dimensional honeycomb lattice of carbon atoms. It is one of the toughest, lightest, and most conductive materials known. Graphene was first isolated using adhesive tape in 2004, and awarded the Physics Nobel Prize in 2010. Here we focus on the mechanical properties of graphene. First we present an analytical study, together with numerical simulations, of ripples in graphene. We show that ripples observed in free-standing graphene sheets can be a consequence of adsorbed OH molecules sitting on random sites. The adsorbates cause the bonds between carbon atoms to lengthen slightly. In the second part of this work we study the fracture mechanics of graphene. Experiments on free-standing graphene can expose the graphene sheets to out-of-plane forces. Here we show that out-of-plane forces can cause free-standing graphene to fracture. This fracture mode is known as the tearing mode and is common in materials such as paper. We present a numerical study of the propagation of cracks in clamped, free-standing graphene as a function of the out-of-plane force. We also obtain an analytical expression for the minimum force required to tear a two dimensional sheet, which is our model of graphene, in terms of the initial crack length.