Capacitive deionisation (CDI) is a desalination technology that removes ionic species from water through electrostatic attraction. Recently, a new CDI cell architecture has been developed with multiple channels and electrodes. Here, the ions are driven to migrate from channel to channel to desalinate the water. The inner electrodes of such a cell function as selective membranes when charged, therefore called capacitive membrane electrodes (CME). This thesis used an experimental approach to investigate the effect of the electrostatic charge and the microstructure of the CMEs on the permselectivity of these CMEs. Five different CMEs were tested in a cell to determine their permselectivities. A porometer (utilising the capillary flow technique) was used to determine the pore size distributions of the CMEs. When comparing the data from both measurements, it was found that CMEs with a higher fluid permeability also showed lower ionic resistances. A lower ionic resistance indicates a lower permselectivity. To achieve good desalination performance in multi-channel CDI, CMEs must offer low ionic resistance when ions are driven from low salinity to high salinity, while still being permselective when ions want to migrate back from high to low salinity.

Enhancing the efficiency of industrial water electrolysis for hydrogen production is vital for the energy transition. In Alkaline Water Electrolysis (AWE), hydrogen is produced at the cathode, and the bubbles are formed when the local hydrogen concentration exceeds the solubility limit. It is important to understand the exact local conditions that result in the nucleation of bubbles in this multi-phase and reactive system. With modeling, it is possible to gain insight into the relation between various local properties, but the model needs to include all relevant physics and chemistry. Thus, this work focuses on the multi-species electrochemical transport phenomena with reaction occurring on the electrode-electrolyte interface.

The electrochemical transport phenomena and the bubble nucleation are meso-scale phenomena occurring at the electrode-electrolyte interface. Lattice Boltzmann Method (LBM) is well suited for modeling meso- scale behavior but it is computationally memory expensive. Consequently, a hybrid approach combining Finite Difference Method (FDM) and LBM has been developed to simulate transport phenomena in the migration-diffusion problem with heterogeneous reaction kinetics. The Debye-Hückel theory is used as a benchmark to validate the developed model. Subsequently, the model is employed to simulate the transport phenomena occurring in the hydrogen half-cell of AWE, with a specific focus on the Hydrogen Evolution Reaction (HER) governed by the Butler-Volmer kinetics equation.

The model captures the dynamic evolution of physical parameters such as electric potential, concentration of species, and fluxes within the system particularly in the Electric-Double layer (EDL). The effect of electrode potential on the distribution of species involved in the reaction are studied by performing simulations for different electrode potential. The influence of secondary fluxes on the species distribution
is studied by implementing a spatially varying boundary condition to the reacting site. Finally, the formulated methodology is extended to solve a multi-phase system with species transportation occurring
around a catalyst particle.

Fresh water is a scarce source. Demand for freshwater is greater than supply, due to the increasing industrialisation of countries and the low natural supply of freshwater sources such as rivers and lakes. Contamination of these natural freshwater sources by industrial wastewater adds to the freshwater deficit. Throughout history, technologies have been developed to increase the supply of fresh water by desalinating other water sources into fresh water. Reverse osmosis is the most widely used desalination technology for seawater desalination, but the energy transition calls for electric-based technologies that can desalinate water using renewable energy sources. Capacitive deionisation is a relatively new desalination technology that uses electrical energy to desalinate brackish water. The capacitive deionisation cell is an electrochemical cell in which the feed water flows between two electrodes. When electricity is applied to these electrodes, ions in the water are attracted to move towards the porous electrodes where they are adsorbed within the electrical double layer. Unfortunately, the capacitive deionisation system has some challenges: it is a batch-system, has a low freshwater production and it has not been built for large-scale production. Avsalt AB, a Swedish company, has developed a new capacitive deionisation architecture to address these challenges, called the multichannel capacitive deionisation system (MC-CDI). This system consists of two capacitive porous electrodes and several membrane-like electrodes between them, referred to as capacitive membrane electrodes (CMEs). To maximise the performance of the multichannel deionisation system, the performance of the capacitive membrane electrodes should be enhanced. Therefore, the aim of his thesis is to optimise the physical properties of the capacitive membrane electrodes that affect the performance of the capacitive membrane electrodes and to gain a better understanding of how the microscopic properties of the CMEs should be carefully tailored to improve the performance.
To address these questions we can turn to Electrochemical Impedance Spectroscopy (EIS). EIS is a non-invasive measurement technique that may be regarded as a much more sophisticated resistance measurement compared to, for example, a multimeter. In contrast to the latter device, EIS measures the impedance, which is a combination of the resistance and the reactance, at a wide range of frequencies. The frequency dependency of the measured impedance can be used to find a so-called equivalent electrical circuit model (EECM). For EIS measurements on electrochemical cells, such as batteries, fuel cells, and electrolysers, the EECM elucidates the different electrochemical processes that occur at different timescales and impedance plots are used to analyse and compare these electrochemical processes. EIS has been successfully used to analyse the electrochemical processes within CDI electrodes and ion exchange membranes, but capacitive membrane electrodes used in an MC-CDI system have never been studied with an accurate and fast measurement technique such as EIS. Because CMEs do not have a solid support structure, in contrast to most CDI electrodes, ions are free to migrate through the electrodes. Therefore, the alternating current that is needed in EIS measurements can be applied on an external set of electrodes, while the response alternating voltage can be picked up at the CMEs. This 4-point impedance measurement configuration enables precise determination of the ionic resistance and capacitance of the electrode material. Therefore, an electrochemical impedance spectroscopy setup is built to determine the performance indicators of the CMEs and to gain a better understanding of the electrochemical processes taking place at the interface of the CMEs. The resulting Nyquist and Bode plots are used to analyse the ion diffusion and capacitive behaviour of the electrodes. To enable the analysis, first, an equivalent electrical circuit is determined for these freestanding electrodes. It was found that the CME can be represented by the Transmission-line model, which in the equivalent electrical circuit takes the form of a junction of three complex impedances. From the values of the electrochemical parameters, the performance indicators of the CME, membrane conductivity and permselectivity, were evaluated.
The CME meter presents itself as an accurate measurement system to quickly evaluate the performance indicators of the CME, with the aim of efficiently searching for the CME that will show the best performance within the MC-CDI system, without placing it within the MC-CDI system. Further research should be conducted to investigate the extent to which EIS could be used to estimate the permselectivity of the CME and whether the total measurement time to predict permselectivity is still short enough to propose EIS as an alternative measurement technique.

In this thesis, a proof of concept was established for the use of a novel coupled QM-MD approach to modelling metallic (copper) electrode-electrolyte interfaces. SCC-DFTB calculations of the instantaneous electronic structure of a copper electrode were coupled to a classical MD simulation of an electrode-electrolyte interface. The applied QM-MD method was described rigorously, and used to investigate the compound distribution and dynamics at the interface, relative to a fully classical MD simulation. Polarisation effects were observed to bring about a significant increase in the attraction between cations and the cathode. Moreover, local polarisation of the cathode was found to immobilise adsorbed cations, and induce an increased orientational preference of the nearby water dipoles. The secondary goal of this thesis was to explore to what extent neural networks are able to replicate SCC-DFTB calculations of the electronic charge density on a metallic electrode. Using a computer vision approach, qualitative evidence was obtained indicating that neural networks can be used to replicate SCC-DFTB predictions on periodic metallic surfaces.

Experimentally investigating the nanoscale behavior at oxide-electrolyte interfaces has proven to be extremely challenging. Molecular Dynamics (MD) simulations have arisen as a potential computational alternative to gain atomic level insights at these interfaces. But how accurately do these simulations represent the physics and chemistry at the interface? In many situations we do in fact not know. Validation at the interface remains challenging. The force fields used in MD simulations, that describe the inter-particle interactions, are generally optimized for purposes deviating considerably from interfaces. Yet, these same force fields are blindly used to model surface-fluid interactions, yielding wildly varying results of for example ion adsorption. This dissertation tackles the problem of simulating interfaces by critically looking at MD simulations and proposing novel solutions, both for MD simulations in general and specifically targeting their validity and limitations with regards to modeling interfaces… @en

This thesis aims to find out whether electrokinetic transport of a calcium chloride solution can be controlled by adding sodium ions by performing molecular dynamics simulations. The idea that electrokinetic transport control is a possibility originates from research that found that charge inversion is reduced when monovalent ions are added to a multivalent solution. Consequently, flow reversal suppression is expected. Many mechanisms are known to contribute to charge inversion and ion competition, but it is unclear how they exactly relate to charge inversion reduction and flow reversal suppression. This work aims to replicate charge inversion reduction and flow reversal suppression in a system with an amorphous silica interface for mixed electrolytes with calcium, sodium and chloride. To allow insight into the structural behaviour of the electrical double layer and the electrokinetic properties for varying concentrations, molecular dynamics simula- tions are used. No variation in charge inversion and flow reversal reduction was found for the simulated concentrations. However, the adsorption behaviour of ions in the electrical double layer changed due to ion competition between sodium and calcium, where sodium outcompetes calcium for inner sphere surface complex adsorption. This work also shows that the electroosmotic flow behaviour for mixtures is sensitive to the dynamic adsorption behaviour of ions, emphasizing the importance of correctly tuned force field parameters between surface and ions.

Force field is widely used to model the potential energy in atomistic simulation systems. Despite force fields have a concise mathematical form, a good set of force field parameters usually requires extra care of calibration. Besides, numerous ionic force field parameters are reported from various sources as researchers have specific target properties for their interests. Previous studies mainly used brute force optimization to find the most desired set of parameters in ionic solution. However, these methods are not efficient since the evaluation of the performance of a parameter set is time-consuming. This work used a stochastic optimization routine in machine learning to tackle the problem of black-box function optimization. This method shows excellent performance of locating the optimum regions of the black-box cost function in only a few iterations. To evaluate the performance of a set of ionic force field parameters, MD simulations are carried out in LAMMPS to compute ionic properties. The solvation free energy and ion oxygen distance are selected as the primary targets while the self-diffusion coefficient and contact ion pairs are regarded as the secondary targets. The optimum region of primary targets are found by direct optimization, then secondary targets are studied with optimized parameters of the primary targets. There have been found discrepancies between the optimum regions of different targeted properties. The dependence studies of individual ionic force field parameters ($\epsilon, \sigma, q$) are analyzed and parameterization trends are found out. Base on these trends, the final calibration model is proposed.

In the pursuit of gaining a better understanding of the mechanisms of oxide-electrolyte interfaces, this thesis presents a working model that mimics a dynamic surface charge distribution by introducing protonation and deprotonation events using MD simulation. Due to the limitations of measurement equipment that operate on an atomic scale, literature cannot provide us with exact time scales for protonation and deprotonation events. Consequently, previous research simulated the surface charge distribution of oxide surfaces as being static and assumed the effect of local protonation and deprotonation to be negligible. This work shows that varying the (de)protonation event period τ significantly influences the characteristics of the electric double layer (EDL). Continuous protonation and deprotonation changes the diffusion coefficient and subsequently alters the structure of the Stern layer, screening function, and preferential adsorption type. As a whole, dynamic surface charge distribution has a considerable impact on the characteristics of the electric double layer depending on τ and should be considered in future MD simulations.