PhD Defence Timon Rijnaarts

the role of membranes in the use of natural salinity gradients for reverse electrodialysis 

Timon Rijnaarts is a PhD student in the research group Membrane Science & Technology. His supervisor is prof.dr.ir. D.C. Nijmeijer from the faculty of Science and Technology (TNW)

There is an urgent need to limit our CO₂ emissions to keep global warming within acceptable levels. To achieve this, the use of renewable energies needs to show a major increase. One of these renewable energies is reverse electrodialysis (RED), where energy is harvested from a salinity gradient. For RED, two streams with a difference in salinity are used in combination with ion exchange membranes. These charge-selective membranes allow either the transport of cations (for cation exchange membranes, CEMs) or anions (for anion exchange membranes, AEMs). Using these membranes, the salinity gradient can be used to facilitate directional charge transport that can be converted into electrical energy. Anticipated feed streams for the salinity gradient can be natural waters, for example, river and seawater. However, the contents of these waters pose challenges to the process. Divalent ions, such as Mg²⁺ and Ca²⁺, have several negative effects (which will be discussed in Chapter 2 – 4). Moreover, dissolved organic matter such as humic acids, is expected to foul AEMs (Chapter 6). Furthermore, the use of spacers in these stacks has drawbacks (such as partially blocking the membrane surface for ion transport) and can be replaced by using profiled membranes. Their design is discussed in Chapter 5.

In previous work, it has been shown that divalent ions have two negative effects in RED. Firstly, when present in the river water, these divalent ions exchange with monovalent ions in the seawater, causing a lower salinity gradient. Secondly, the resistance of the stack increases when divalent ions are present. In Chapter 2, these effects of divalent cations in RED are investigated using CEMs with different divalent transport properties. Monovalent-selective CEMs are used to block divalent cations, while multivalent-permeable CEMs are selected to allow the transport of divalent cations. These membranes are characterized on their monovalent-ion selectivity using membrane resistance in NaCl and MgCl₂, and show large differences in the transport of divalent cations. In RED, monovalent-selective CEMs are able to block divalent cations transport and are able to operate without voltage losses due to a lack of uphill transport. Alternatively, multivalent-permeable CEMs allow divalent cation transport and do not experience a membrane resistance increase. Both these specific types of CEMs improve gross power densities by at least 30% compared to standard-grade CEMs when divalent cations are present in the feed water.

In Chapter 3, uphill transport by divalent cations from the river water to the seawater is discussed. This uphill transport causes a voltage loss, and is still present with the improvements gained in Chapter 2 using multivalent-permeable CEMs. Here, a pretreatment using Donnan Dialysis (DD) is proposed to exchange divalent cations from the river water with monovalent cations in the seawater to avoid uphill transport. Surprisingly, we do not find an improvement in the obtained voltage in RED for DD-pretreated river water with divalent cations. This is caused by the exchange of divalent for monovalent cations in DD prior to the RED process, where the effective salinity gradient is lowered because divalent cations in the river water are exchanged by Na⁺ in the seawater. However, this exchange of Na⁺ to the river water decreases the resistance of the river water – which comprises over 80% of the total stack resistance. This leads to an overall improvement of 6.3% in net power density.

In Chapter 4, tunable monovalent-selective membranes are made using polyelectrolyte multilayer coatings on commercial ion exchange membranes. After the promising results for monovalent-selective membranes discussed in Chapter 2, it is important to devise a tunable and simple method to prepare such membranes. We find that for these multilayer coatings a net positive charge and a low swelling ratio, in combination with a defect-free multilayer are important properties for monovalent-selective coatings. The coated membranes have high monovalent cation selectivity (of Na⁺/Mg²⁺) up to 7.8 compared to 3.5 for the uncoated CEM, for experiments on single membranes. In large-scale electrodialysis stack experiments, this leads to an improved monovalent cation selectivity of 5.4 compared to 1.4 for uncoated CEMs. However, these multilayers coating did not enhance monovalent-selectivity for AEMs, likely due to the positive charge of the multilayer.These tunable coatings could be used to counteract uphill transport in RED as well.

In Chapter 5, profiled membranes are compared with using spacers in RED. These profiled membranes do not block part of the membrane surface nor the channel for ion transport, which is the case for spacers (the so-called spacer shadow effect). In this work, pillar-profiled and chevron-profiled membranes are compared to RED with spacers. It is found that profiled-membranes indeed have an improved stack resistance due to absence of the spacer shadow effect. Chevron-profiled membranes improve the mixing as well, leading to even lower resistances at the cost, however, of additional pressure drops in the channels. In the end, chevron-profiled membranes are found to have the highest net power densities of all studied configurations. However, alignment of the chevron profiles during assembly of the RED stack proves difficult and, therefore, an improved design without alignment issues is proposed based on crossed chevrons.

In Chapter 6, the effect of natural water fouling in RED is discussed, with a specific focus on the role of AEMs. Therefore, six different AEMs were tested in RED stacks using natural waters at the Afsluitdijk and these AEMs were separately characterized on the membrane level before and after the RED experiments. It is found that the change of properties of the AEMs (permselectivity and membrane resistance) due to fouling in natural waters depends on the chemistry and water content of the membranes. For the RED experiments, a decrease of 15 – 20% in power density is observed for all AEMs. However, the change in AEM properties due to fouling can only explain 2 – 4% decreases in power density. The largest effect of fouling losses in power density are assumed to be caused by spacer fouling.

In Chapter 7, a discussion on the findings in this thesis is given. Based on these results, three main challenges for future research on RED are proposed. Firstly, the role of divalent anions is not studied separately as has been done for divalent cations. When present in the river water, these are expected to cause uphill transport, just like divalent cations do. Secondly, natural fouling should be studied on different profiled membranes, to discuss the influence of profile shape on fouling tendency. Finally, to enable investigation of different profile shapes, rapid prototyping using 3D printing can be used. Alternative uses of RED are discussed as well, where the salinity gradient is used to drive other electrochemical reactions than those required to produce electricity.