Understanding Transport Phenomena in Membrane Systems for Waste Utilisation: Electrodialysis Concepts for Waste Heat to Hydrogen and Lithium-Ion Battery Recycling

Abstract

The overall aim of this thesis is the development of useful and sustainable applications of ion-exchange membranes, with a focus on methods for understanding the fundamental transport processes taking place in the electrochemical cell. A deeper insight into the fundamentals may aid in the design and development of both new ion-exchange membranes and process applications. This approach is applied to the modeling of electrodialysis and reverse electrodialysis for use in battery recycling and hydrogen production, respectively. First, models for the ion-exchange membrane fluxes and permselectivity were derived for an aqueous KCl mixture using non-equilibrium thermodynamics. The established relations were used to model the state of a unit cell during steady state reverse electrodialysis operation, and a stack of unit cells, each with an electric potential contribution due to the salinity gradient, was used to drive water electrolysis. The spent electrolyte mixtures exiting each unit cells could subsequently be regenerated by distillation using industrial waste heat, restoring the salinity gradient and creating a closed-loop system for hydrogen production. Simulations indicated a hydrogen production of 0.38 gH2 m−2h−1, a waste heat requirement of 1.7 kWh g−1 H2 , and a system efficiency of 2 % is possible with a saturated feed mixture and a draw mixture of 0.2 mol kg−1. The permselectivity model showed that the transference of water has a significant impact on the unit cell electric potential and therefore also on the system performance. Furthermore, it depends on the magnitude of the coupling between electric current and water flux, but also on the thermodynamic state of the two electrolyte mixtures. Factors such as the water transference coefficient, membrane and mixture resistance, cell geometry and the choice of electrolyte have been identified as important parameters which can be optimised to improve the system performance. Next the permselectivity model was verified experimentally for two commercial ionexchange membranes, the Selemion CMVN cation-exchange membrane (CEM) and the Selemion AMVN anion-exchange membrane (AEM). It was determined that the model could accurately describe the statistically significant trends in the electric potential measurements of ion-exchange membranes subject to a salinity gradient. It was also verified that the electric potential contribution of one membrane was more easily accessed by using bare Ag/AgCl electrodes compared to commercial reference electrodes with filling solutions and porous junctions. The porous junction of the reference electrode showed an electric potential contribution to the total voltage compatible with a K+ transport number of tK+ = 0.494 ± 0.008 and no significant net transference of water. This is congruent with literature values of measured transport numbers for K+ in bulk aqueous KCl. The Selemion CMVN CEM was characterised by tK+ = 0.996 ± 0.006 and a constant water transference coefficient of tw = 3.69 ± 0.40. The Selemion AMVN was well described by tK+ = −0.002±0.004 and tw = −3.75±0.27, and therefore tCl– = 1−tK+ = 1.002±0.004. In other words, the two membranes appear to be perfectly selective to the target ion, but each ion carries around 4 water molecules through the membrane as part of the charge migration process. This is shown to have a beneficial effect on the energy requirement of electrodialysis, but a negative effect on the available work from a reverse electrodialysis cell. The permselectivity model was then extended to ternary mixtures of KCl, H2O and ethanol (EtOH). A computational model was created to explore the state of a unit cell in electrodialysis, for the purpose of extracting salt and water from a saturated feed mixture and leaving purified EtOH behind. To do so, the ionexchange membranes must be highly selective towards salt along with a significant amount of water co-transport, while restricting co-transport of EtOH. In a case of individual ion-exchange membranes with tw = 15 and ta = 0, the EtOH solvent weight fraction was increased from ωa,in = 0.7 to ωa,out = 0.81. This required around 73 kWh m−3 EtOH, and could serve as the basis for a salt precipitation process with an energy requirement of 0.161 kWh mol−1 KCl. The concept may be of use in the recycling of batteries if ion-exchange membranes with negligible EtOH co-transport are identified or developed in the future. For the analysis of measurements and construction of models related to the ternary mixture, thermodynamic data is necessary. The collected data was combined and critically analysed with a focus on the reference state of the salt. Notably, a constant reference state was used, in contrast to the variable salt reference state often used for mixed solvent electrolyte mixtures. Empirical regression models were then formulated for the activity coefficients of H2O and EtOH, and KCl was subsequently expressed as a function of those two models via Gibbs Duhem’s equation. This procedure yielded a thermodynamically consistent model which could be regarded as a useful low-complexity mixed solvent electrolyte model. Lastly, we extended the permselectivity measurements to the ternary mixture for the determination of ion-exchange membrane transference coefficients. Significant variance was present in the measurements, but the permselectivity of the Selemion CMVN cation-exchange membrane was well described by a constant tK+ = 0.98 ± 0.01, a constant water transference coefficient of tw = 2.81 ± 0.42. The EtOH transference was characterized by the ratio of the transference coefficient to the thermodynamic activity, ta/aa, which was determined as a function of the mean thermodynamic activities of EtOH and salt in combination with a constant regression coefficient χ = 1.38 ± 0.40. EtOH transference coefficients of up to 2 were calculated based on this model. These results suggest that up to 2 molecules of EtOH can be carried along with K+ as the electric current passes through the CEM. The established permselectivity expressions and experimental methods are general and can be extended to other mixtures of interest.