Long-term durability is becoming increasingly relevant for capacitive deionization (CDI) as the technology emerges on the commercial scale. Short-term deionization studies have suggested that Faradaic leakages could be a major factor in electrode degradation, but the long-term effects are still unclear. In this study, we probe the degradation process of the desalination efficiency in commercial CDI modules for up to 52 days of non-stop operation. This corresponds to a little more than 100 m3 of water treated, and the lifetime production volume of the modules is estimated between 150,000–250,000 L of purified water. Surprisingly, the results demonstrate that the absolute long-term loss is largely linear with the cumulative charge leakage. This suggests short-term leakage currents could reasonably predict long-term degradation. Interestingly, the absolute loss mechanisms mean devices with higher total capacitance are more degradation resistant. Finally, shortening cycle times and other methods of reducing leakage would lead to a proportionally longer lifetime. Notably, the first 2 min of the 10 min operation retained 50 % of the performance with only 10 % of the leakage (10-fold reduction). In conclusion, the work provides a method for understanding, predicting, and reducing degradation in long-term operations with commercial CDI modules.

Capacitive deionization (CDI) is an emerging technology in water cleaning applications. Bipolar electrode modules are promising for upscaling CDI systems by applying a high voltage over a stack of cells. Previous experiments indicate around 3 times faster removal rate can be achieved with a 5-cell stack of flow-between electrodes. In this work, we present the first flow-through bipolar CDI design. Surprisingly, the effective average salt adsorption rate (ASAR) of a 5-cell stack was around 10 times higher per electrode mass. The flow-through design allows for faster ion transport to match the higher charging rate. Also, the floating electrodes discharge near-instantaneous by internal short-circuit upon the removal of potential, leaving almost twice as much time for charging in a cycle. Resistance is lowered due to the removal of internal compartments in the design, and the benefits of flow-through devices outweigh any potential higher energy cost of mass transport. The high voltage at the extreme electrodes poses a risk of electrode oxidation but optimized device operation controls this risk by constraining the voltage locally. For the same reason, the device is best adapted for the deionization of moderate to low ionic concentrations, such as river water, well water, or municipal water. In summary, this work provides a simpler yet highly effective way of scaling up CDI for water-cleaning applications.

Electrochemical deionization devices are crucial for meeting global freshwater demands. One such is capacitive deionization (CDI), which is an emerging technology especially suited for brackish water desalination. In this work, we extend an electrolytic capacitor (ELC) model that exploits the similarities between CDI systems and supercapacitor/battery systems. Compared to the previous work, we introduce new implementational strategies for enhanced stability, a more detailed method of describing charge efficiency, layered integration of leakage reactions, and theory extensions to new material and operational conditions. Thanks to the stability and flexibility the approach brings, the current work can present the first fully coupled and spatiotemporal three-dimensional (3D) CDI model. We hope that this can pave the way toward generalized and full-scale modeling of CDI units under varying conditions. A 3D model can be beneficial for investigating asymmetric CDI device structures, and the work investigates a flow-through device structure with inlet and outlet pipes at the center and corners, respectively. The results show that dead (low-flow) areas can reduce desalination rates while also raising the total leakage. However, the ionic flux in this device is still enough under normal operating conditions to ensure reasonable performance. In conclusion, researchers will now have some flexibility in designing device structures that are not perfectly symmetric (real-life case), and hence we share the model files to facilitate future research with 3D modeling of these electrochemical deionization devices.

This research aims to explore the impact of tilted channel configurations of CDI cells on desalination performance. The results reveal that the titled convergent channels have a faster average salt adsorption rate (ASAR) than the regular straight geometry. For desalination operations that end at a quarter of the equilibrium salt adsorption capacity (SAC), the convergent spacer with a slight slope of 1.5 degrees has a 20 % higher ASAR than the typical straight geometry (0.15 mg/g/min for convergent and 0.12 mg/g/min for straight). This gain increases to about 24, 29.5, and 33%, respectively, for slopes of 3.5, 5.5, and 7 degrees, compared to the straight geometry with the same spacer thickness. By looking at the underlying mechanisms, the spacer geometry is found to shift the location of the initial adsorption. This affects how quickly the device outputs the cleaned water. Interestingly, the geometry angle can also affect the location of the depletion zone, so tilted spacers can also affect the behavior during electrode starvation. Specifically, the convergent geometry has the depletion zone in the middle of the electrode instead of the corner near the outlet, as seen for straight and divergent channels. Together, these findings indicate how to construct tilted spacers to enhance CDI performance.

Intercalation host compounds (IHC) are promising for selective ion removal from water. Recent theoretical developments have suggested that electrochemical desalination with IHC (nickel hexacyanoferrate (NiHCF)) electrodes could separate K⁺ and Na⁺ by a factor of 160. However, the experiments only produce a selectivity of around 3. In this work, we derive theory and a finite-element (FEM) model to investigate the origins of time-dependent selectivity suppression. The first results show that ion starvation can severely limit selectivity. Surprisingly, we also find that operations at low state-of-charge produce theoretical selectivity of 600, which is way above what was previously thought to be the theoretical maximum. Also surprising is that the main cause of low selectivity is that the constant-current overpotential disproportionally favors the adsorption of the ion that is less selected in the equilibrium state. By implementing short charging cycles near the depleted state with rest periods at the ends, we raised the time-dependent selectivity from 3 to 450. Even higher output selectivity could be achieved by combining IHC cathodes with membrane-less carbon anodes. In conclusion, the insights and methods derived here could enable highly selective ion removal at the device level for a wide class of IHC materials.

Capacitive deionization (CDI) is a desalination method that has been expanding substantially in recent years. As processes are getting more complex, corresponding developments in theory and software are necessary to keep up and drive future research. In this work, we derive a new CDI theory based on a tertiary current distribution, meaning each ionic species is resolved individually in a unified framework in 1D/2D. The results show that this approach is ideal for simulations with multiple ionic species and materials that affect cations and anions differently. Direct examples include such as intercalation materials and membranes. It is also effective for incorporating electrode replenishment in flow-electrode CDI (FCDI). By benchmarking with traditional methods, we demonstrate that numerical stability is a central limitation of traditional methods for these applications. The results identify physical processes involving rapid changes to cause major instabilities. This can thus be handled by introducing specific numerically stabilizing factors. Finally, the theory is compiled into comprehensive software that researchers can straightforwardly apply in future studies without having to reconstruct methods from scratch. A corresponding video tutorial has also been deposited. In conclusion, the work pushes the limits of the simulation capabilities in a wide range of CDI processes.

Capacitive deionization (CDI) is an emerging technique for purifying water by removing ions. Recent experimental studies have reported that the anion/cation adsorption can be naturally imbalanced, even for a solution with just sodium and chloride, and suggested a link between imbalance and Faradaic leakages. However, these effects have been missing from conventional models. In this work, we developed a new circuit model to better understand the connection between Faradaic leakages and adsorption imbalance. The theory demonstrates that the effect emerges in a model that includes leakages, considers leakages on both electrodes separately, and considers different leakage resistance on the two electrodes. Having the model, it is possible to analyze and quantify the influence of the leakage resistance and other material properties on the adsorption imbalance. Leveraging these results, we further present a multi-electrode (ME) device design. The setup adds a third electrode to the spacer channel and can tune or eliminate the adsorption imbalance based on appropriately distributing the voltage across the electrodes. In conclusion, we describe a charge leakage mechanism responsible for the imbalance of ion adsorption and a flexible device design to tune the anion/cation removal.

Clean water is a major global challenge. Meanwhile, capacitive deionization (CDI) is an emerging desalination technology that could help produce and reuse water. As the technology develops, the modeling of upscaled systems is becoming increasingly relevant. However, the inherent complexities in the CDI process have historically made such simulations unfeasible. In this work, we leverage the newly published electrolytic-capacitor (ECL) model to efficiently simulate parallel/serial flow modes in CDI stacks. The simulations are based on finite-element methods (FEM) that couple differential equations for describing local charging and ionic transport inside the device. The results show that both parallel and serial connections scale incredibly well with the system size. Still, parallel connections have the advantage of requiring lower pumping energy. Overall, we find that the relationship between adsorption capacity, flowrate, and compartment size is a good indicator of performance. In conclusion, the ELC model is promising for simulating upscaled CDI.

The massive need for freshwater is driving new desalination technologies such as capacitive deionization (CDI), wherein an applied electric field between porous electrodes removes salt ions from water. In this work, we present substantial advances in numerical approaches to 2D finite-element models that make it possible to tractably and accurately simulate the local transport, charge-transfer, and ion-adsorption processes. This is achieved by introducing a new numerical approach that improves the stability of the method (SmD), which further allows precise and effective modeling that was previously too unstable for use in the state-of-the-art 2D models. The results show that the model now accurately and reliably simulates CDI processes while being effectively applicable to a wider range of structural (device level) and operational conditions (like flow). Crucially, this opens up new opportunities that allow us to provide novel insights into the CDI processes, especially relating to ion-starved conditions. Finally, novel algorithms support fully automatic implementation with simultaneous fit to multiple data sets and we openly provide all software code to increase accessibility. Thus, we fundamentally believe that the developed model will provide a solid foundation for 2D spatiotemporal simulations of capacitive deionization and aid the future development of CDI technology.

More than 2 billion people are living in water-scarce areas. Meanwhile, there are enormous amounts of water in the salty oceans. Capacitive deionization (CDI) rises to the challenge with electrochemical cells for desalinating the water. As the CDI field expands, modeling becomes an increasingly important part of the development landscape. Multiscale modeling could bring innovations from the material scale to pilot plants. 

The multiscale work in this thesis has been like climbing a mountain. At the start, we investigate the macroscopic device level. One milestone is the electrolytic-capacitor (ELC) model, which can simulate CDI process dynamics. Whereas previous 2D models were unsteady for a single CDIcell, the ELC model could accurately simulate stacks of over 100 cells at a fraction of the time. It also enables simulations of complex upscaled geometries, such as bipolar electrode stacks, ohmic charging, and asymmetric devices. Going up the mountain, the mesoscopic level reveals the local mechanisms behind the macroscopic behavior. One important stepping stone is the dynamic Langmuir (DL) model, which reveals how isotherm-based modeling can crease stable and tractable simulations. Also, developments in isotherm, double-layer, and circuit modeling make it possible to choose what model structures to lean on depending on the conditions. Near the top of the mountain, the microscopic level shows the fundamental atomic mechanisms behind the mesoscopic material properties. These investigations reveal a ladder mechanism of ion transport in crystals of Prussian blue analogs (PBA), meaning the cations climb frames formed by negative groups in the crystal structure.

In the end, we plant a flag by combining the developments from the journey into a complete multiscale model. That model demonstrated that we could predict CDI charging trends from the atomic structure of PBA electrodes. Having the full multiscale model also made it possible to backtrack and determine atomic-level mechanisms by comparing the output of different mechanism cases with macroscopic experiment data. The multiscale mountain is massive and has big potential. A dream is that future research will expand these concepts, in CDI and beyond.

Över 2 miljarder människor lever i dag i områden med vattenbrist, samtidigt som det finns enorma mängder saltvatten i haven. Kapacitiv avjonisering (CDI) kan hantera detta genom avsaltning av vatten med hjälp av elektrokemiska celler. När CDI-fältet expanderar blir också modellering allt viktigare. Speciellt med multiskalemodellering finns möjligheten att driva innovationer från material till pilotanläggningar. 

Vårt jobb har varit som att klättra upp för ett berg. I den inledande delen undersökte vi den makroskopiska nivån, som handlar om hur avsaltningsenheterna fungerar. Ett viktigt steg för att simulera dynamiken i processen har varit utvecklingen av ELC modellen. Till skillnad från tidigare modeller som kunde vara instabila för en enda avsaltningscell så kunde ELC-modellen hantera travar med över 100 celler. Det gör det möjligt att simulera komplexa uppskalade strukturer, såsom bipolära elektroder, ohmsk laddning, och asymmetrisk design. Vidare upp i berget finns mesoskalan. Den visar på de lokala mekanismerna bakom det makroskopiska beteendet. En viktig del har varit den dynamiska Langmuir-modellen (DL), som har visat hur isotermbaserad modellering kan ge stabila och smidiga simuleringar. Utvecklingen i isoterm-, dubbellager-, och kretsmodeller gör det även möjligt att välja lämpliga metoder att stödja sig mot beroende på situation. Nära toppen av berget finns mikroskalan, som handlar om det atomära beteendet som bestämmer de mesoskopiska egenskaperna. Här har vi upptäckt en stegmekanism för jontransport i kristaller av berlinerblått. Detta innebär att katjoner klättar längs ramar som utgörs av negativa grupper i kristallstrukturen.

Slutligen hissar vi flaggan genom att kombinera resultaten från alla nivåer. Multiskalemodellen visar att vi kan förutsäga laddningstrender i CDI baserat på atomstrukturen i elektroden. Multiskalemodellen gjorde det också möjligt att gå baklänges och att identifiera mekanismer på mikroskala genom att beräkna den makroskopiska effekten av olika fall och jämföra med experimentella data. Multiskaleberget är massivt och har stor potential. En dröm är att framtida forskning ska utöka koncepten från den här avhandlingen, i CDI och vidare.