In this thesis, I have used computational methods to the study of flexibility in metal-organic frameworks (MOFs) and the behavior of adsorbed gases and liquids in MOFs. Among nanoporous materials, MOFs are of particular interest for adsorption-based applications for reasons that include (1) their huge chemical space, which raises the possibility of tuning them for specific industrial processes and (2) the ability of the under-coordinated metals present in some MOFs to endow their respective materials with impressive adsorbate binding energies.

Chapters 2 and 4 relate to analogs of a series of MOFs with under-coordinated metals (the M-MOF-74 series). In Chapter 2, I report the discovery of a novel argon adsorbate-induced deformation pattern for this framework series. This result was arrived at using a flexible framework model, and is presented as an explanation for an intriguing signal observed in experimental small-angle X-ray scattering profiles upon argon adsorption. This hypothesis is supported by a complementary investigation of an alternative explanation for the X-ray signal, in which argon atoms are proposed to adsorb at different densities in adjacent pores.

In Chapter 3, I studied the dynamics of adsorbed xylene isomers in MOF-5. I compare the translational and rotational motion of the xylene isomers, and explain the differences based on molecular geometries. The most rod-like molecule, para-xylene, is more rotationally constricted in the pore due to its intermolecular interactions with the aromatic group of the MOF linker. This finding has implications for the rational design of MOFs, as a process that can take advantage of the MOF-induced variations in xylene isomer dynamics could be used for lucrative liquid-phase xylene separations.

In Chapter 4, I present a method for parameterizing the type of flexible framework model used in Chapter 2 from ab initio quantum chemistry calculations. The goal of this work is to facilitate the rapid development of flexible, versatile MOF models that can capture changes in the coordination bonding of metals in MOFs. This type of model can be used to study structural transitions and may be extended to study MOF formation. We demonstrated that our approach yields models for the M-MOF-74 series that are stable and have simulated structural properties in good agreement with quantum chemistry calculations.

In this thesis, I investigate nanoporous materials such as metal-organic frameworks (MOFs) and covalent organic frameworks (COFs) for various gas adsorption applications using a wide array of computational methods. These types of materials are ideal for gas adsorption and separation applications due to their large internal surface areas and tunable chemistry. They are also ideally suited to study using traditional computational methods due to their well-defined structures. In the first chapter, I introduce nanoporous materials and the various molecular mechanics methods which can be used to study them. I also introduce the topic of in silico materials design. Then, in the next chapter, I discuss the development of a DFT-derived force field to accurately study the gas adsorption behavior in materials which contain coordinatively unsaturated metal sites. In such materials, the most commonly used methods fail to accurately model adsorption behavior, and the introduction of the DFT-derived force field has allowed the study of flue-gas mixtures in these frameworks. Following this work, in the third chapter we discuss the use of the DFT-derived force field to study the dynamical behavior of greenhouse gases in the same MOF series. Much of this work was done in collaboration with experimentalists who used NMR as their primary tool to probe the dynamics of these gases in the materials. Our molecular dynamics simulations complemented their NMR experiments. In the fourth chapter, I switch gears and discuss the use of computational methods for the design of new materials, first to characterize experimentally synthesized materials, and then to construct a database of thousands of new COF structures. Finally, I conclude by sharing a summary of my findings from the various investigations discussed in this thesis and my future outlook for the field.

Microporous materials are a diverse class of materials that includes porous carbons, zeolites, metal-organic frameworks (MOFs), and zeolite-templated carbons (ZTCs). These materials are characterized by a high surface area and diversity of pore geometry, as well as chemical tunability in MOFs, which provides a large design space for applications in gas storage and separations, catalysis, energy storage, sensing, and medicine.

The first chapter of this dissertation introduces microporous materials, their applications, and methods for studying them. In Chapter 2, I discuss a study that investigated the potential to tune the product distribution of the propene dimerization reaction via shape selectivity in MOFs and zeolites. We showed that the shape of the pore indeed exerts a large influence on the distribution of branched and linear products from propene dimerization. We demonstrated, moreover, that it is possible to predict the experimental linear conversion through the use of Monte Carlo simulations to compute the contribution of the heat of adsorption of the different products to their free energies of formation. Using this approach, we screened a library of MOFs and zeolites and identified materials which could further increase the linear conversion of propene dimerization beyond the highest-performing materials identified in the literature.

In the next chapter, Chapter 3, I detail our investigation into the adsorption of Δ9-tetrahydrocannabinol (THC) in MOFs, in order to find a material that can be used in a THC breathalyzer. We tested two models of THC, one with only the bulky, rigid head group, and another describing the full molecule, including the flexible alkyl tail; we found that the simplified THC-head model is sufficient in most cases to rank the adsorption energy and Henry coefficient of materials. We also compare NVT and Widom insertion methods for computing the heat of adsorption of the materials, finding that in general the NVT method provides more statistically reliable values from shorter simulation times than Widom insertions. We introduced a novel method to evaluate the pore accessibility to nonspherical adsorbates, based on calculating the overlap area between the pore and an ellipsoidal probe molecule, and used this criterion to further narrow the pool of promising candidates. From our screening, we identified three motifs promoting high THC adsorption, and highlight the highest-performing MOFs exhibiting these motifs. Finally, we computed the Henry coefficients of water in our materials and determined a selectivity threshold that would enable the detection of THC from humid breath.

In Chapter 4, I present results from reactive force field simulations of a set of amine-appended Mg2(dobpdc) MOFs, which have shown promising performance for carbon capture. We studied the mobility of the the amine functional groups in the MOFs, and found that they undergo a hopping motion in which the ends of the amine swap their coordination bond with the metal center of the MOF. We computed the rates of amine hopping and the activation enthalpies and entropies of the transition, and found that it is possible to tune them by using functional groups with more or less steric hindrance at the amine ends, giving us insight into how different functional groups affect the step temperature and pressure of CO2 adsorption.

In Chapter 5, I discuss the performance of ZTCs as the electrode material in an electrical-double layer capacitor (EDLC). We built EDLC simulation cells from a library of ZTC materials and investigated multiple equilibration protocols, determining that equilibrating with a constant-potential simulation is the most suitable choice. We then showed that the charging rate is strongly correlated with the pore size of the materials, which suggests that ion diffusion within the pores is the limiting process for electrode charging. Unlike the charging rate, the equilibrium capacitance of the materials was not clearly correlated to common geometric descriptors, but we found that it was strongly correlated to the charge compensation per carbon (CCpC), which indicates the efficiency of charge storage in the materials. In order to better understand the charge storage mechanisms in ZTCs, we examined the local charges and ion adsorption sites of individual pores in the materials, and identified characteristics of pores with higher and lower than average CCpC. We found that the CCpC is enhanced for adsorption sites near pockets with a small radius of curvature, and suggest this as a guideline for the design of improved EDLC electrode materials.

The final chapter concludes the dissertation and presents some perspectives on the outlook of these results.

Nanoporous materials, such as metal-organic frameworks (MOFs), have enormous internal surface areas. Their consequent adsorption properties demonstrate promise towards solving energy-related problems in gas storage and gas separations. Owing to their modular and versatile chemistry, millions of possible nanoporous materials can be synthesized. This vast chemical space allows a material to be tailor-made or fine-tuned to target specific adsorbate molecules and conditions. In this thesis, we utilize molecular models and simulations of gas adsorption in both existing and predicted nanoporous material structures to accelerate the discovery of new materials targeted for gas storage and separations at specific conditions.

In the first part of this work, we approach the problem of identifying an optimal porous material to densify natural gas for storage onboard vehicles as fuel. We developed a series of statistical mechanical models to find the thermodynamic parameters that optimize the deliverable capacity of a material. We conclude that the heat of adsorption, which is a commonly used metric to evaluate materials for natural gas storage, is a misleading metric because the optimal heat of adsorption depends on the pore size. Our models also reveal that adsorbate-adsorbate attractions-- in the case where multiple methane molecules can fit into a pore-- can enhance the deliverable capacity.

Next, we carried out a high-throughput computational screening of metal-organic frameworks, porous polymer networks, zeolites, and zeolitic imidazolate frameworks for natural gas storage. The data that we collected provide candidate structures for synthesis, reveal relationships between structural characteristics and performance, and suggest that it may be difficult to reach the current Advanced Research Project Agency-Energy (ARPA-E) deliverable capacity target. To assess thermodynamic limits to the methane deliverable capacity, we then built a model of an extreme scenario where an energy field can be created without taking up space with material. This model suggests that, while the failure to reach the ARPA-E storage target is due to material design constraints rather than purely thermodynamic constraints, the ARPA-E storage target is ambitiously close to the thermodynamic limit.

In the second part of this work, we approach the problem of identifying a material that selectively adsorbs xenon over krypton. With over half a million nanoporous material structures to consider as candidate adsorbents, the computational cost of a brute-force computational screening strategy was prohibitive. Instead, we employed a machine learning algorithm, a random forest, to learn the relationship between quickly computed structural descriptors and Xe/Kr selectivity, which is more expensive to compute. The trained random forest allowed us to rule out a large percentage of the materials on the basis of quickly-computed structural descriptors. Our machine learning accelerated screening pinpoints top candidates on which to focus experimental efforts and elucidates structure-property relationships for design guidelines for a Xe-selective material.

As we are now working with mixed gas adsorption, we developed a user-friendly software package in Python, pyIAST, for ideal adsorbed solution theory (IAST) calculations. IAST is a thermodynamic framework to predict mixed gas adsorption from pure-component adsorption isotherms, which are easier to measure. We provide practical guidelines for applying IAST.

Finally, we carry out a high-throughput computational screening of metal-organic frameworks for capturing Xe from air at dilute conditions, a separation encountered in used nuclear fuel reprocessing. Our computational screening, facilitated by a parallelized code on GPUs, predicted a metal-organic framework, SBMOF-1, to be among the most Xe-selective. Our experimental collaborators synthesized and tested SBMOF-1 and found it to exhibit the highest Xe/Kr selectivity and Xe Henry coefficient reported in the literature. Column-breakthrough experiments reveal that SBMOF-1 is a near-term material for capturing xenon from the off-gases of used nuclear fuel reprocessing plants. This is a rare case of a computation-assisted materials discovery.

Lipid bilayers, along with embedded inclusions such as cholesterol and proteins, constitute a biological membrane—the interface between a cell or organelle and its environment. Understanding the structure of a biological membrane and the physical principles responsible for the organization of membrane inclusions is crucial to understanding the processes that occur on the surface of a cell or organelle such as inter-cellular signaling, immune synapse processes, exo- and endocytosis, and membrane fusion. In this work, we study the organization and effect of inclusions in a lipid bilayer using statistical mechanics principles and large-scale molecular simulations.

In the first part of this work, we propose a generic physical force for the assembly of inclusions in lipid bilayers—the orderphobic effect. Inspired by modern theories of the hydrophobic effect, this force is governed by the physics of the first-order phase transition between the ordered (i.e., solid-like) and disordered (i.e., fluid) phases of lipid bilayers. We show the existence and nature of this force using coarse-grained molecular dynamics simulations, and demonstrate the lateral assembly of two model proteins, or ‘orderphobes’, within a lipid bilayer. The effect is powerful and operates at nano- to mesoscopic scales, and could be a potential explanation for the clustering of proteins in a biological membrane.

In the second part of this work, we focus on hydrophobic inclusions within the lipid bilayer that give rise to organelles known as lipid droplets. These droplets, earlier thought to be passive stores of hydrophobic material in an otherwise aqueous cytosol, have now been implicated in various metabolic diseases. Since the growth and behavior of nascent droplets is sub-microscopic, little is known about the details of this process. We construct a computational framework that allows for a lipid reservoir to study asymmetric inclusions in a lipid bilayer such as the lipid droplet. Further, we identify key stages in the growth and budding of this organelle.

In this work we use mesoscopic simulation to investigate the short and medium range interaction of transmembrane helices. First, we develop a coarse-grained model of a generic transmembrane α-helix. We use the geometry constraints of an α-helix, along with a set

of homogeneous hydrophobic beads. We discuss the reasoning behind this modeling and why various other plausible models have failed in describing the short-range interactions of transmembrane helices.

We then show the effect of hydrophobic mismatch on these model helices. The tilt angle of a single helix is affected by its hydrophobic mismatch with the surrounding bilayer. We show that the hydrophobic mismatch, via its effect on the tilt angle, has a crucial effect on the cross angle of two packed transmembrane helices as well.

We further show the effect of hydrophobic mismatch on the potential of mean force between the helices as well as on the thickness and tilt angle of the lipids surrounding the helices. We introduce a class of especially long helices, termed super-positive mismatched

helices, to which the response of the lipids differs and discuss the origin of this behavior.

These coarse-grained simulations are performed using the Dissipative Particle Dynamics (DPD) scheme Performing large time and length scale simulations using this scheme is highly CPU-time consuming. To enhance our simulation capabilities we develop a novel, massively parallel simulation algorithm. We introduce this algorithm and the original concepts we developed for parallelizing the DPD scheme. We show that this novel approach provides

up to 30 times speed up on a Graphical Processing Unit (GPU) over the non-parallel CPU version.

Finally, we use our parallel algorithm to perform simulation on a large time and length scale system. We investigate the effect of Cholesterol on the bending rigidity of lipid bilayers and show the effect of a phase transition on the bending modulus. Extracting the bending modulus from simulation requires a large system that enables long-range height fluctuations. Using a fast, parallel simulation algorithm for such a task is therefore crucial.

Since tens of thousands of nanoporous materials have been synthesized and millions have been hypothesized, our ability to both predict novel materials and their performance for a given application in a computational setting is becoming increasingly important. Utilizing simulations that can screen for optimal material performance becomes invaluable since exhaustive experimental exploration of this vast chemical space can quickly become experimentally intractable from an expense and time efficiency standpoint. This thesis is therefore dedicated to advancing the computational tools and algorithms that can be used to both predict novel materials and evaluate their performance for clean energy technologies.

Nanoporous materials such as metal-organic frameworks (MOFs) are promising for applications in clean energy thanks to their diverse and extraordinary gas adsorption properties. Additionally, these materials' high degree of tunability means that nearly infinite distinct materials can be envisioned with different combinations of compositions, topologies, and surface properties. Each combination could yield a material with its own unique gas adsorption phenomena and resulting real-world applications. While many materials with outstanding properties have been studied, it remains a challenge to find or design an optimal material for a given application and to understand the complex molecular-level interactions that give rise to macroscopic behavior.

In this dissertation, we take three different computational approaches towards advancing the discovery and understanding of nanoporous materials for various clean energy applications.

In Chapters 2 and 3, we screen large databases of materials for carbon capture and hydrogen storage, using molecular simulations. We develop accurate atomistic models for the relevant adsorption phenomena and identify appropriate criteria for predicting performance. In this way, we identify optimal materials for carbon capture and hydrogen storage, as well as characteristics common among top performers. Ultimately, such findings can help guide the design of optimal materials for these applications.

In Chapter 4, we study in depth a novel biporous MOF for natural gas processing, using molecular simulations in conjunction with experiments. This MOF's rare feature of biporosity, or coexistence of two chemically distinct pores within the same MOF, gives rise to unique stepwise activation that can be controlled to consequently tune the MOF's CO2/CH4 separation performance. This phenomenon is explained in terms of CO2's and CH4's differing interactions with the pore surfaces. In this way, we contribute to the understanding of complex behavior in biporous materials that affects their gas separation performance.

Finally, in Chapter 5, we investigate the assembly process of MOFs using a statistical mechanical lattice model. In particular, we map out the thermodynamic assembly pathway of MOF-74 and explore the effect of surface tension energies on the shapes of nucleating clusters. With this study we work towards understanding the assembly pathway of MOFs with the ultimate goal of being able to control the synthesis process and obtain crystals of the desired morphology.