Grenfell Patey

Pure water does not freeze homogeneously at temperatures above -35°C. Heterogeneous ice nucleation induced by ice nucleating particles (INPs) occurs at warmer temperatures than homogeneous nucleation. Atmospherically relevant INPs can significantly affect ice formation in mixed-phase and ice clouds, and therefore influence the Earth’s climate system. Several factors, including inorganic ions, acidic or basic solutes, and pressure variations influence the ice nucleating ability of INPs. At present, little is known about the microscopic mechanisms by which these factors affect heterogeneous ice nucleation and the ice nucleating ability of INPs.We examine the influences of inorganic ions and pH changes on kaolinite’s ice nucleating ability employing direct molecular dynamics (MD) simulations. We demonstrate that ionic solutes decrease the ice nucleating ability of the Al (001) kaolinite surface. The hydroxy groups exposed on the Al (001) surface can become deprotonated in basic solution or dual-protonated in acidic solution. We find that the mono-protonated Al (001) surface, which is expected to be stable at near-neutral pH, is the most effective ice nucleating surface. We perform droplet freezing experiments with kaolinite samples under acidic and basic conditions, covering a pH range of 0.18 - 13.26. Our MD simulations and experimental results indicate that the Al (001) surface may be important for ice nucleation by kaolinite.Employing MD simulations and varying the protonation state of the α-alumina (0001) surface, we find that the ice nucleating efficiency of this surface decreases nearly symmetrically with acidic and basic pH changes away from the neutral point. This is in qualitative agreement with previous droplet freezing experiments performed on a single (0001) surface. We carry out droplet freezing experiments on two α-alumina powder samples, covering a pH range of 0.59 - 13.19. The freezing results for the powder differ significantly from those of the single surface.We use MD simulations to probe the influence of pressure on heterogeneous ice nucleation by a finite-size β-AgI disk. We find that negative pressures enhance ice nucleation at high temperature, and we demonstrate how water diffusion can significantly influence ice nucleation and growth at low temperatures and negative pressures.

View record

The lithium halide salts may at first appear to be simple chemical systems. However, previous research indicates lithium halides have complex and poorly understood crystallization pathways from aqueous solutions. While lithium halides exist in the rocksalt crystal structure under ambient conditions, common lithium halide classical force fields more often predict wurtzite as the stable structure. This failure severely limits their application in molecular simulations of crystallization. Research in this thesis focuses on presenting new results and computational methodology to better understand the structural chemistry of both lithium halides and alkali halides in general.Density functional theory together with classical force fields are employed to examine the relative stability of candidate crystal structures for lithium halides and produce accurate theoretical reference data. Dispersion interactions are shown to play a key role in the stability of rocksalt over closely competing crystal structures. Classical models can be corrected in their structural predictions by scaling up the strength of dispersion, indicating a pathway towards better lithium halide force fields.Convolutional neural networks are used for the structural classification of simulated alkali halides. The neural networks are trained on a large data set generated from molecular dynamics simulations of alkali halides across a range of temperatures. Time convolution filters out short-lived structural fluctuations. The structure classifiers are shown to be accurate in bulk phase simulations, then demonstrated on crystallization of model alkali halide systems from the melt. The neural network classifiers are implemented in a melting point calculation algorithm for model binary salts. Finite size effects are characterized, then melting points of alkali halides using common rigid-ion interaction potentials are calculated and discussed.The methods developed throughout the research are employed for the optimization of pairwise lithium halide force fields, fitted to reference data using a reinforcement learning approach, Bayesian optimization. Limitations on the Coulomb Lennard-Jones potential form are uncovered, which do not appear to hold for the more flexible Coulomb Buckingham potential form.By introducing advanced computational methodology, this research reveals the inherent structural complexity of lithium halides and emphasizes the importance of considering structural landscapes during classical forcefields construction for molecular simulation.

View record

Heterogeneous ice nucleation (HIN) has applications in climate science, nanotechnology, andcryopreservation. Pure water does not freeze homogeneously above ∼ −39◦C. Ice nucleationon the Earth’s surface or in the atmosphere usually occurs heterogeneously involving foreignsubstrates (ice nuclei). Foreign substrates can be mineral dust, soot, or bacteria. Experimentsidentify good ice nuclei but lack sufficient microscopic resolution to answer the basic question:What makes a good ice nucleus? We employ molecular dynamics (MD) simulations to addressthis question and investigate microscopic mechanisms of HIN. We consider substrates that display a wide range of ice-nucleating abilities. These areα-alumina, kaolinite, gibbsite, hematite, mica, β-AgI, and K-feldspar. The α-alumina (0001)surface nucleates the (0001) basal plane of hexagonal ice (Ih), whereas the kaolinite (001) planestabilizes the (10¯10) prism face. Mica (001) surfaces nucleate ice by an unusual mechanism, theice nucleus consists of ordered layers of hexagonal and cubic ice. The primary prism face (10¯10)of β-AgI nucleates the prism face of Ih, which significantly enhances current understandingof HIN by AgI. All substrates mentioned above are good ice nuclei in experiments. HIN isnot observed for gibbsite, hematite, and boehmite, again consistent with their experimentalclassification as poor ice nuclei. Experimentally, K-feldspar is an excellent ice nucleus, butour simulations show no icelike order for its (001), (010), and (100) planes. This suggests thatsome surface rearrangement or defect is responsible for HIN by K-feldspar. Simulations of multiple surfaces allow us to identify important features of a good icenucleus. We develop a two-dimensional (2D) lattice description of water ordering in thesurface layer. Basal and prism faces ice bilayers decompose into two, 2D lattices (triangularfor basal, rectangular for prism), and a substrate must stabilize both to promote HIN. We demonstrate how lattice match, hydrogen bonding, and atomistic surface morphology cruciallyinfluence HIN efficiency. We employ machine learning to predict the likelihood of HIN using surface features, togetherwith liquid water properties local to the surface, which can be obtained from shortsimulations. This achieves 89% accuracy in the classification of surfaces as good or bad icenuclei, with water properties the dominant factor.

View record

In this dissertation, we investigate water and ion transport through carbon nanotubes usingmolecular dynamics simulations. Specifically, we examine how different water models influence the simulated conduction rates. We consider three common water models, which areTIP4P/2005, SPC/E, and TIP3P, and observe that water flow rates through the same nanotube are strikingly different amongst the different water models. Also, the water flow ratedependence on temperature fits an Arrhenius-type equation over a temperature range from260 to 320 K. We provide evidence that there are two factors which determine the conductionrate: the bulk fluid mobility, and the molecular structure of confined water. For narrow nanotubes, for example, a (6,6) nanotube, where water only forms a single-file configuration, thefirst factor can largely account for the flow rate differences. In this case, we show that the conduction rate correlates with the diffusion coefficient of bulk water. Our simulation results arewell described by continuum hydrodynamics as well. The factor of bulk fluid mobility is stillimportant in the water conduction through intermediate-size nanotubes, such as a (9,9) nan-otube. Also, the formation of complex configurations within such nanotubes can impede thetransport rate by influencing the mode of water conduction. The ordered structure occurringwithin nanotubes can also explain the differences between simulation results and continuumhydrodynamics predictions. Hence, both factors decide the water conduction rates throughintermediate-size nanotubes. Moreover, we demonstrate that the ion flow rate depends on theviscosity of the bulk solution, as well as the water structure within the nanotubes, togetherwith the ion size. In particular, at lower temperatures complex water configurations act toimpede ion transport while still allowing water to flow at a significant rate. In general, ourefforts on this issue are of importance for future simulation studies investigating water and ion conduction through nanoscopic channels. This dissertation might also prove useful indesigning more efficient nanoscopic conduits for future experimental studies.

View record

The process of crystal nucleation, despite being so fundamental and ubiquitous in industrial and natural processes, is still not fully understood because of its stochastic nature, and the high spatial and temporal resolution needed to observe it through experiments. This thesis investigates several aspects of nucleation through the use of molecular dynamics, a computational technique that is able to simulate systems up to ~10¹² atoms (as of today's computational power).The projects in this thesis focus on the nucleation from aqueous solution of alkali halide salts, with supplementary studies on the related processes of dissolution in water, and crystallization from the melt.The mechanism of NaCl nucleation from solution is examined in Chapter 3 by direct simulation. The NaCl supersaturated solution was found to contain many small ionic clusters that continuously form and disappear from solution until one (or more) of them nucleates and grows irreversibly. An original method was developed to detect and follow clusters in time, producing results useful in the study of their characteristics and lifetimes. Most importantly, it was found that the lifetime of transient clusters is about ~1 ns, and that both the cluster lifetime and nucleation probability are significantly higher if the cluster is more geometrically ordered. The dissolution of NaCl crystals was also investigated. The process was found to happen in stages, is characterized by an activation barrier, and can be described by a simple rate law. The crystal nucleation of LiF from supersaturated solution was observed, in our simulations, only at high pressure and temperature. The growth rate for an already nucleated crystal was found to have a temperature dependence that follows the Arrhenius law, and further evidence suggests that the reason for such behavior is the high activation energy required to dehydrate the ions.The crystallization from the melt of the Joung-Cheatham and Tosi-Fumi models for lithium halides was also investigated. We found that, for the Tosi-Fumi model, all lithium halides crystallize as wurtzite. For the Joung-Cheatham model, LiF and LiCl crystallize as rock salt, while LiBr and LiI crystallize as wurtzite.

View record

Water droplets in the atmosphere do not freeze homogeneously until -38ºC. Freezing at warmer temperatures requires heterogeneous ice nuclei (IN). Despite the importance of ice in the atmosphere, little is known about the microscopic mechanisms of heterogeneous ice nucleation. This thesis employs molecular dynamics simulations to investigate ice nucleation by silver iodide, kaolinite, potassium feldspar, gibbsite, and a protein. Silver iodide is one of the best known ice nucleating agents. We examined seven surfaces of silver iodide and observed ice nucleation on three surfaces. The surfaces that nucleated ice organized the first layer of water molecules into a configuration resembling ice, such as chair conformed hexagonal rings. Surfaces that do not nucleate ice do not organize water into icelike configurations, such as planar rings. Results suggest lattice mismatch is insufficient in predicting ice nucleation, and a finer atomistic match is required. Finite silver iodide disks and plates were used to probe the relationship between the size of a nucleating surface and maximum temperature of ice nucleation. Larger disks nucleated ice at warmer temperatures than smaller disks by forming larger initial cluster of ice which could reach the critical size easier than homogeneously formed clusters. Kaolinite is a common clay known to nucleate ice. Our simulations investigated both sides of the (001) surface and found both sides able to nucleate ice. The Al-surface was simulated with varying degrees of freedom of motion. An optimum amount of movement was required to nucleate ice as the surface needs to adapt into a configuration favorable to ice. Ice nucleated on the Si-surface via the formation of a novel composite surface structure which facilitated bulk ice nucleation. Potassium feldspar simulations explored three variations of the two primary cleavage planes. All surfaces failed to nucleate ice and density profiles suggest that the surfaces are unlikely to nucleate ice. We succeeded in nucleating ice on gibbsite with prepared surface conformations compatible with ice. Biological IN, such as ice nucleation proteins, are among the most efficient IN. We attempted to simulate ice nucleation via a protein, but were unable to achieve ice nucleation.

View record

We examine the melting point trends across sets of coarse grain model salts using NPT molecular dynamics simulations. The melting point trends are established relative to a charge-centered, size-symmetric salt that is closely akin to the restricted primitive model. Two of the common features of ionic liquids, namely size asymmetry and a distributed cation charge, are systematically varied in a set of model salts. We find that redistributing the cation charge in salts with size-symmetric, monovalent, spherical ions can reduce the melting temperature by up to 50% compared to the charge-centered case. Displacing the charge from the ion center reduces the enthalpy of the liquid more than that of the solid resulting in a lower melting point. We consider two sets of size-asymmetric salts with size ratios up to 3:1 using different length scales; the melting point trends are different in each set, but within each set we find salts that achieve a melting point reduction of over 60% from the charge-centered, size-symmetric case. The lowest melting point range we find is between 450 K and 500 K. We find diversity in the solid phase structures. For all size ratios with small cation charge displacements, the salts crystallize with orientationally disordered cations. For equal-sized ions, once the cation charge is moved far enough off-center, the salts become trapped in glassy states upon cooling and we find an underlying crystal structure (space group 111) that features orientationally ordered ion pairs. The salts with large size ratios and large cation charge displacements achieve the lowest melting points and also show premelting transitions at lower temperatures (two as low as 300 K). We find two types of premelting behaviour; some salts exhibit a fast ion conductor phase, where the smaller anions move through a face-centered cubic (fcc) cation lattice, whereas other salts have a plastic crystal phase composed of ion pairs rotating on an fcc lattice.

View record

This thesis aims to understand the influence of electric fields on ice nucleation. Molecular dynamics simulations are employed to investigate heterogeneous ice nucleation induced by electric fields, and why external electric fields promote freezing in liquid water models.The first project considers heterogeneous ice nucleation in systems, where water molecules experience an electric field in a narrow region over an entire surface. The specific focus is ice nucleation and growth processes. Different water models are considered, and the influences of temperature and field parameters are examined. We find no qualitative difference between the two water models. By analyzing structure, we show that a ferroelectric cubic ice layer freezes inside the field region, and unpolarized ice grows beyond the field region, at temperatures not far below the melting point. We explore ice nucleation by electric field bands, which act only over a portion of a surface. Field bands of different geometry nucleate ice, provided that the band is sufficiently large. Analysis of different systems reveals that ice strongly prefers to grow at the (111) crystal plane of cubic ice, and that ice nucleated by field bands usually grows as a mixture of cubic and hexagonal ice. Our results suggest that local electric fields could play a major role in heterogeneous ice nucleation, particularly for rough particles with many surface structural variations, that serve as ice nuclei in the environment. We also investigate the electrofreezing of water subject to a uniform field. The aim is to obtain an understanding of why electric fields facilitate ice nucleation. It is shown that the melting point of water increases significantly when water is polarized by a field. The increased melting point is mainly due to the favourable interaction of near perfectly polarized cubic ice with the applied field. Relevant to the mechanism of heterogeneous ice nucleation by local surface fields, our results suggest that local fields effectively increase the degree of supercooling of locally polarized liquid. This decreases the size of the critical nucleus in the region influenced by the field, facilitating ice nucleation.

View record

Grand canonical Monte Carlo calculations are used to determine water adsorption and structure on kaolinite surfaces, with and without the presence of trench-like defects, as a function of relative humidity (RH), at 235 K and 298 K. Both basal planes (the Al- and Si-surfaces), as well as two edge-like, defect free surfaces are considered. The trenches simulated are rectangular in geometry, and have a fixed depth and varying width. Thegeneral force field CLAYFF is used together with the SPC/E and TIP5P-E models for water. At both 235 K and 298 K, the edges, Al-surface, and trenches adsorb water at sub-saturation, in the atmospherically relevant pressure range. The Si-surface remains dry up to saturation. Both edges and the Al-surface adsorb water up to monolayercoverage. Adsorption on the Al-surface exhibits properties of a first-order process with evidence of collective behavior, whereas adsorption on the edges is essentially continuous and appears dominated by strong water lattice interactions. Only next to the Al-surface, were hexagonal rings observed in the water layer. However, they did not match hexagonal ice Ih. The results obtained using trenches show that the granularity of the surfaces can play a major role in the adsorption of multiple layers of water over a large range of RH. Our calculations suggest that water adsorption in trenches, and possibly in other similar defects, can offer an explanation of the large water coverages reported experimentally. Related to ice, the very dense, proton ordered, ferroelectric structures found in the trenches at235 K do not correspond to any recognizable form of bulk ice. We speculate how these structures might aid ice nucleation and growth, and suggest how this possibility could be further explored with simulations and experiments.

View record

Room temperature ionic liquids (RTILs) are pure, organic salts, that are liquid at ambient temperatures. RTILs are highly customizable, with a wide choice of ion types and varying substituents. The customizability of RTILs also poses the greatest challenge, rational design of liquid properties from molecular structure.The influence of RTIL ion characteristics on liquid structure and transport properties is systematically investigated employing molecular dynamics simulations. The characteristics investigated include size disparity, ion charges that are displaced from the center of mass, and variation of the cation-anion charge separation σ'(+−). Different simple spherical ionic liquid models are developed that isolate each of these characteristics.Statistical mechanical analysis shows that more size disparate models have decreased coordination numbers, the diffusion coefficients and electrical conductivity increase, and the viscosity decreases. The effects of size disparity can be canceled to a large extent by decreasing σ'(+−). An increasing displacement of the ion charge leads to the formation of directional ion pairs of increasing strength. Weak pairing results in non-uniform ion distributions and reduced caging, with similar liquid property trends as size disparate systems. A large charge displacement and short σ'(+−) leads to increasing numbers of strong, long-lived directional ion pairs that dominate the liquid behavior. Increasing viscosities and decreasing electrical conductivities are observed in this regime. The temperature behavior of the ionic liquid models deviates from linear Arrhenius behavior, especially for the conductivity. The relationship between diffusion and viscosity conforms to the fractional Stokes-Einstein equation.The qualitative conclusions of our calculations suggest the utilization of ions with moderate charge displacement, and large size disparities, for desired low-viscosity RTILs with large ion mobilities.Mixtures of water with ionic liquid models generally show increased diffusion coefficients and electrical conductivities, and decreased viscosities. The increased mobility of the ions can be mainly ascribed to dynamical effects due to light water replacing heavy ions in the ion coordination shell. We observe deviating behavior when water can form strong, directional interactions with at least one ion, in the case of small, or of strongly charge displaced ions. In both cases the viscosity increases with increasing water concentration.

View record