Surface modification of water filtration membrane materials has the potential to imbue attractive qualities, such as enhanced antifouling or solute rejection/selectivity capabilities. Development of next-generation membrane materials will therefore rely on design rules relating surface structure and functional properties. Heuristics exist for simple surfaces comprising homogeneous chemistries and have largely been discovered through trial-and-error synthesis and characterization. However, we lack a complete understanding of nanoscopic mechanisms governing the water-mediated interactions between contaminants/foulants and realistic surfaces with chemical and topological heterogeneity. Furthermore, trial-and-error methods are not able to efficiently explore the large design space offered by multifunctional surfaces.

This dissertation develops relationships between surface chemistry and local water/solute behavior for model membrane materials, using computational inverse design approaches. By coupling molecular dynamics simulations with optimization algorithms, we identify non-intuitive chemistries showing enhanced functional properties. Genetic algorithm optimization of patterns of nonpolar and polar chemistries in a model nanopore suggests that arrangement of polar chemistry in rings along the nanopore wall will enhance selectivity of water over small, neutral solutes. Active-learning-based optimization of more complex patterns with both chemical and topological heterogeneity suggest that rings of large nonpolar groups create a rough surface that achieves similar effects. Importantly, these non-intuitive functionalities discovered by the inverse design approach suggest previously unexplored mechanisms governing water-mediated solute transport in model membrane materials.

Sequence-specific polymers such as polypeptoids provide a synthetically accessible route to chemically precise surface modification. However, exploration of sequence-structure-function relationships for polypeptoids has been lacking, in part because computational workflows to simulate polypeptoids are not well established. In collaboration with detailed experimental characterization, we validate an enhanced sampling simulation workflow and atomistic model for polypeptoids. We then show that polypeptoid sequence modulates local water behavior in dilute solution and at peptoid brush surfaces, and that the changes in water structure can be related to trends in affinity of small-molecule solutes for peptoid brush surfaces. Finally, we apply the previously developed inverse design techniques to build sequence-structure relationships by identifying sequences with extremal conformational behaviors.

In this thesis, we outline the development of a multiscale physics-based platform for exploring and ultimately predicting the design-specific interaction of ~1-10 nm particles with model cellular membranes. Nanoparticles (NP) are ever-present in foods and beverages, cosmetics, packaging, cooking products, fertilizers, pesticides, and novel pharmaceuticals, and pose significant challenges related to their increased consumer, occupational, and environmental exposure and their unique bioactivity relative to small molecules and large colloids. Thanks to rapidly advancing fabrication and characterization techniques, NPs are highly tunable in physicochemical properties such as size, surface chemistry, shape, elasticity, roughness, and crystallinity. Currently, however, the influence of these NP design parameters is highly underdeveloped, and difficult to reproducibly demonstrate in in vivo, in vitro, and even model experiments. Specifically, NP interactions with and passive transport across cellular membranes play a significant role in pharmacological and consumer product performance (biodistribution) as well as adverse outcome pathways in toxicology. We thus focus on the fundamental problem of design-specific interactions between NPs and cellular membranes, modeled to a first approximation as lipid bilayers.

To provide accurate, efficient, and robust predictions for a range of NP designs, we construct a first-of-its kind, multiscale physics-based platform linking detailed molecular dynamics (MD) simulations, continuum mechanical theory, and multi-compartment modeling. Using this platform, we examine the two most influential design parameters--size and surface chemistry--and through two main case studies: (1) the membrane permeability of sub-nanometer particles and (2) the thermodynamic stability of larger-scale, ~1-10 nm particle-membrane interactions. Within (1), we first simulate the NP-membrane interactions and transport in full detail to test the validity of Overton's Rule, a longstanding structure-property relationship, and the inhomogeneous solubility-diffusion (ISD) model, a microscopic mechanistic continuum model for transport. We show that Overton's Rule is overly simplified for describing transport across a fluctuating lipid bilayer membrane, yet that ISD model holds for small enough particles. Within this range of particles where the solubility-diffusion mechanism holds, we directly link the impact of particle chemistry in the MD simulations to transient (time-dependent) transport outcomes in the macroscopic multi-compartmental models. This allows us to both compare with and evaluate models used in experimental permeability assays and close the orders of magnitude gap between simulation-predicted and experimentally-calculated permeabilities. We also leverage our platform to construct improved structure-property relationships for the steady-state membrane permeability and structure-kinetic relationships, accounting for a wider range of particle chemistries and highlighting the imperative of time in dictating the design rules for membrane permeation.

Within case study (2), we probe larger NP-membrane interactions that implicate macroscopic membrane deformations and restructuring. Using the molecular simulations, we map out in particle size and chemistry space the putative NP-membrane interaction configurations, some of which resemble and agree with small-scale solubility-diffusion theory or large-scale membrane elastic theory (e.g. for lipid bilayer or monolayer wrapping of the NP) in stability limits and free energies and some of which require closer examination in the simulation to explain the thermodynamics. We also discover an entirely novel mechanism of interaction for ~4 nm, rough crystalline hydrophobic particles that we call "asymmetric leaflet hopping," wherein the particle preferentially inserts in one bilayer leaflet, forming a pre-pore in the membrane and inducing large-scale membrane curvature, and flips to the other leaflet over extended time scales.

We conclude with preliminary phenomenologies to outline the phase behavior of ~1-10 nm particles of varying chemistry, as well as other areas where our platform shows great promise. By accounting for a vast range of NP designs, natural lipid diversity (lipidomics), and variable compartmental size, boundary layer, and transient conditions, this platform has the potential to more intuitively and effectively inform systems-level physiologically-based pharmacokinetic models for NP biodistribution predictions, as well as structure-activity relationships for direct predictions of product efficacy and toxicity. The end result of this multiscale platform is that we can directly link a NP's microscopic physicochemical properties to its macroscopic outcomes in a dynamic biodistribution setting.

High-throughput density functional theory calculations can provide a method for predicting material properties in alkali-ion electrode and electrolyte chemistries. Through these calculations, paired with cluster expansion models and uncertainty quantification methods, we have tools to understand the fundamental thermodynamic origins of crystallographic deformation and alkali-ion phase stabilities in these materials. In this analysis, we use these methods to understand phase stability and lithiation mechanisms in a family of high power density electrode chemistries called the Wadsley-Roth crystallographic shear phases. Due to the complexity and large cell sizes of Wadsley-Roth phase materials, this combination of methods is useful for tapping the origin of the high charge rates and power densities in these unique structures. The study begins with an examination of stability of compounds in the Ti-Nb-O ternary where we examine the effect of electrostatics and distortions on phase stability of the Wadsley-Roth phases. We then examine the effect of chemical strain in the lithium site-filling mechanism in the highly reversible Wadsley-Roth material, PNb9O25. We extend our analysis to the high-power density, commercialized Wadsley-Roth phase, TiNb2O7, where we examine the effect of metal-metal electronic interactions and pair distances on the complex lithium site filling mechanism present in this compound. We then do an in depth analysis of the effect of lithiation on octahedral distortions to acquire a complete understanding of structural changes upon lithiation. These analyses are paired with experimental results to exhibit the validity of these calculations.