Supported carbon electrode

Technical field

The present invention relates to an electrode suitable for capacitive deionization (CDI), as well as a process for manufacturing an electrode.

Background

Electrochemical devices, such as a batteries, electrolysers, and capacitive deionizers (CDI) comprise electrodes. The electrode constitute the electrical conductor, at which surface the desired electrochemical reactions take place. For example, in a capacitive deionizer for desalination of water, the salts of the feed water are removed by electrosorption, i.e. by adsorbing or absorbing salt ions at the surface of electrical polarised electrodes. The performance or efficiency of a CDI electrode thus relates to the rate of adsorption/absorption, as well as the amount of adsorped/absorped ions, also referred to as the electrode salt absorbance capacity (SAC).

To improve the performance of CDI electrodes, electrode materials with high salt absorbance capacity are used. The SAC is determined by the material as well as the microstructure of the material, since the electrosorption occurs at the surfaces of the electrode. Hence, the electrode material is typically a carbon derived material, and the electrode structure is typically a thin, porous layer, which has a high surface area, and low electrical resistance. For example WO 2009/065023 [1] describes a porous graphite electrode for a deioniser with a thickness below 0.75 mm.

Due to the porosity and layer thickness, the electrode is typically not a robust, free standing structure, and therefore associated with an additional external mechanical support, i.e. located externally to the active electrode. For example, WO 2009/065023 [1] describes a fluid permeable separator placed between the two electrodes, where the separator acts as a mechanical support for the electrodes and further allows the feed water to flow through. The separator is exemplified as non-electron conductive structure, e.g. a glass fiber paper, separating the electrically conducting and active electrodes.

Despite the advances within the field of electrodes, there is a need for improved electrodes and particularly improved electrodes suitable for capacitive deionizers (CDI). Summary

The present disclose provides an improved electrode, particularly for capacitive deionizers, comprising a three-layered structure having an inner layer located, or sandwiched, between two outer layers. The three layers comprise carbon particles and are electrically conducting (i.e. electronic conducting and/or ionic conducting), however, the central, inner layer further comprises a fibrous, scaffold support, providing a mechanical framework. The scaffold support thereby facilitates a self-supporting or free-standing electrode, which may be separately and independently manufactured, stored, and integrated into a system.

A first aspect of the invention relates to an electrode, comprising an inner layer sandwiched between two outer layers, wherein the outer layers comprise multiple first carbon particles, and wherein the inner layer comprises an internal fibrous, scaffold support and multiple second carbon particles distributed within the scaffold support, thereby forming a three-dimensional, percolating network of second carbon particles.

A second aspect of the invention relates to a process for manufacturing an electrode, comprising the steps of: a) mixing at least one polymeric binder with an organic solvent, b) adding one or more dispersants and/or viscosity regulators, and carbon powder to the mixture, c) stirring the mixture to form a flowable slurry, d) providing a fibrous, scaffold support, e) depositing a layer of the slurry on both sides of the porous support, thereby impregnating the fibrous, scaffold support, and f) drying the applied slurry to evaporate the organic solvent.

Description of Drawings The invention will in the following be described in greater detail with reference to the accompanying drawings.

Figure 1 shows a cross sectional view of an embodiment of an electrode according to the present disclosure, where (A) shows an embodiment of the three layers, and (B) shows a close up of an embodiment of the inner layer. Figure 2 shows an embodiment of a system comprising multiple electrodes according to the present disclosure.

Figure 3 shows an embodiment of a CDI system including multiple flow channels. Figure 4 shows an embodiment of a CDI system including multiple flow channels. Figure 5 shows experimental data for the salt concentration in the channels during operation, which is a way of testing the efficiency of the electrodes according to the present disclosure. For example, the electrodes according to Example 1 may be tested in similar experiments.

Figure 6 shows an embodiment of an electrode according to the present disclosure (B), and a conventional single layer electrode (A).

Figure 7 shows a scanning electron image (SEI) of an embodiment of an electrode according to the present disclosure.

Figure 8 shows an embodiment of a CDI device including a field generator configured for four phases.

Figure 9 shows experimental data for the salt concentration in the channels during operation of an embodiment of the electrodes according to Example 3.

Figure 10 shows a scanning electron image of an embodiment of an electrode according to the present disclosure, such as the electrode of Example 5.

Detailed description

The invention is described below with the help of the accompanying figures. It would be appreciated by the people skilled in the art that the same feature or component of the device are referred with the same reference numeral in different figures. A list of the reference numbers can be found at the end of the detailed description section.

Definitions

By the term “support” is meant a mechanical support, which may be either an external support structure or an internal support structure. For example, an active thin film (e.g. an electrode film) may be deposited onto a thicker supporting layer, whereby the supporting layer acts as an external mechanical support for the active thin film. This is particularly useful when the thin film is too fragile to be self-supporting and/or manually handled. Examples of an internal support structure include a continuous backbone, scaffold, or frame forming a network located internally within a layer, which provides mechanical strength to the active layer, thereby enabling the layer to be self-supporting and to be manually handled.

By the term “percolating network” is meant the interconnected pathway formed between randomly dispersed particles. For example, in a mixture comprising spheric particles of carbon packed into a closest packing by gravity, neighboring carbon particles are in contact, and the contact points form a three dimensional interconnection ranging throughout the mixture, e.g. from one surface of the mixture to any opposing surface of the mixture. Similarly, the interstitial pores between the packed carbon particles will form a percolating porosity pathway, such that a liquid may be percolated through the mixture from one side to a second side (e.g. the opposite side) of the packed structure.

By the term “electrosorption” is meant adsorption or absorption of ions at the electrode material. Thus, the term interchangeably covers adsorption of ions at the surface of an electrode material, and absorption of ions into e.g. internal pores within an electrode material.

By the term “bimodal size distribution” is meant a particle size distribution having at least two modes or peaks, corresponding to two different particle size fractions. For example, a mixture of a powder fraction with a finer average particle size, and a powder fraction with a coarser average particle size, will form a bimodal size distribution with a fine and a coarse peak.

Electrode for electrosorption

Capacitive deionization (CDI) devices remove salt from water using porous electrodes that absorb positive and negative ions from the water that flows adjacent to their surfaces. This absorbance can be initiated by applying an electric voltage, e.g. preferably around 1 volt between a pair of electrodes, which will induce migration of positive ions towards the negative electrode and negative ions towards the positive electrode. As a result, the salinity of the water decreases. This electrosorption is reversible, and desorption of the ions may occur when the voltage is reversed. Thus, the performance of a CDI electrode may be related to the salt absorbance or adsorbance capacity (SAC), and the rate (kinetics) at which the ions can be absorbed or adsorbed. Carbon-derived materials, particularly activated carbon (AC), combine high SAC values with good electronic conductivity, enabling the electric polarization, and are conventionally used for making CDI electrodes. Carbon-derived materials also include carbon nanotubes, and graphene flakes. The carbon materials are typically in the form of powders consisting of multiple carbon particles. The powders or particles are typically bonded together using a polymeric binder, and the mixture of carbon particles and polymeric binder may then be shaped into electrodes. The overall shape of the electrode is typically a thin sheet, which enables fast ion migration towards the pores.

The specific structure and microstructure of the electrodes will determine the performance, i.e. the maximum capacitance, salt sorption capacity, and electrical resistance of the electrode. Generally, the electrode material should have a large surface area, high electrical conductivity, good wettability, high mechanical strength and robustness. Increasing electrode resistance results in an impeding effect on the electrosorption, whereas increasing the porosity and electrode thickness or mass facilitate that the electrode has a larger SAC and can adsorb more salt. However, the adsorption kinetics are more limited for electrodes of high mass and possibly higher porosity and tortuosity due to the increase in diffusion time. Also the risk of water cross over from neighbouring channels in a multi-channel CDI increases with increasing porosity and reduced thickness. Hence, improved electrical conductivity may be obtained at the cost of reduced capacitance and SAC,

Further, for multi-channel CDI the electrode materials must facilitate that ions can adsorb or absorp into and permeate the electrode, while simultaneously reducing the water cross-over between neighbouring channels as this reduces the water recovery efficiency, as also described below (cf. the “system” section). Also, for multi-channel CDI the electrode structure is advantageously uniform across the thickness or symmetrical across the thickness, such that the salt adsorption is isotropic. Hence, the electrode behaves similar irrespective of the side the salt approaches the electrode, which is particularly relevant for multi-channel CDI.

Layered structure

Electrodes with a layered structure may provide improved performance, including higher SAC values, and higher ion absorption and migration rates, as well as improved mechanical properties. Further, a layered structure may make an external support structure dispensable, and thereby facilitate a larger active electrode area for ion absorption. Thus, a layered electrode may facilitate ion absorption from all surfaces of the electrode, in contrast to an electrode supported by an external mechanical support, where the surface in contact with the support cannot be active. Figure 1A shows a cross sectional view of an embodiment of an electrode 2 with a layered structure according to the present disclosure. The electrode comprises three layers: a central, inner layer 2.2 sandwiched between two outer layers 2.1 , i.e. an upper outer layer and a lower outer layer.

All three layers comprise carbon or carbon-derived materials, preferably in the form of multiple particles, to facilitate SAC and electrical conductivity. The inner layer 2.2 further comprises an internal fibrous, scaffold support, as seen in the close up of the inner layer in Figure 1 B. The fibers 2.3 are physically connected and thereby form a three-dimensional scaffold, network or backbone, which provides a mechanical framework within the inner layer. The carbon particles 2.4 and 2.5 of the inner layer 2.2 are distributed within the scaffold structure, whereby the carbon particles also form a three-dimensional network of carbon particles. The combination of the fibrous network and the carbon particles facilitate that the carbon particles are distributed such that they form a percolating network, meaning that the carbon particles are connected to form pathways for electrons 7, as indicated by the fine dashed lined arrow in Figure 1 B. Thus, the inner layer facilitates both electrical conductivity and a mechanical support structure.

In an embodiment of the disclosure, the electrode comprises an inner layer sandwiched between two outer layers, wherein the outer layers comprise multiple first carbon or carbon-derived material particles, and wherein the inner layer comprises an internal fibrous, scaffold support and multiple second carbon particles distributed within the scaffold support, thereby forming a three-dimensional, percolating network of second carbon particles.

Advantageously, all three layers further comprise porosity to increase the surface area of the electrode, and thereby the interaction area with the ions to be adsorped or absorbed. Preferably, the layers comprise a three-dimensional, percolating network of pores, thereby forming pathways for the adsorbed ions to migrate within and across the electrode layers. In Figure 1B, the pathways for ions 6 are indicated by the coarse dashed lined arrow. Thus, the layers advantageously in addition facilitate ionic conductivity.

Due to the fibrous, scaffold structure of the inner layer providing sufficient mechanical support, the electrode layers may include a high amount of porosity, providing both high adsorption capacity, high adsorption rates, and high active surface area, while being mechanical robust. Thus, the layered electrode may be provided as a self- supporting or free-standing electrode that is separately and independently manufactured, stored, and integrated into a system. This is particularly advantageous for CDI systems, where the number of electrodes integrated into the system may be easily reduced or increased depending on the feed liquid.

In an embodiment of the disclosure, the electrode is a capacitive deionization (CDI) electrode for electrosorption.

The layered structure may further provide the advantage of an electrode having layers with different properties. For example, the outer layers may be configured for mainly improving the electrochemical performance, whereas the inner layer may be configured for mainly mechanical support, and thus be essentially electrochemically passive, and e.g. mainly provide for chemical diffusion. For example, it was found that if the outer layers or the inner layer contribute with a relatively low electronic resistance by its own, then the gain from a conductive support material is relatively small or even negligible. Therefore, a non-conductive support material can surprisingly be used instead of a conductive support material without expected decrease of conductive performance.

Carbon

To ensure high SAC and electrical conductivity of the electrode, all three layers comprise carbon or carbon-derived materials, and advantageously the carbon is in the form of powder or particles, and comprises carbon types and particle sizes that induce high ion absorbance capacity, as well as sufficiently high amounts of carbon to enable the electric polarization. Hence, the carbon of the outer layers may be referred to as first carbon particles, and the carbon of the inner layer may be referred to as second carbon particles. Advantageously, the inner layer has similar properties as the outer layers, i.e. high SAC and electrical conductivity. Thus, the inner layer advantageously comprises the same carbon types, carbon particle sizes, and carbon amounts, as the outer layers. Hence, preferably the first carbon particles are identical to the second carbon particles.

In an embodiment of the disclosure, the first and second carbon particles are the same.

In a further embodiment, the first and/or second carbon particles are selected from the group of: activated carbon (AC), carbon black (CB), graphite, carbon fibers, and combinations thereof.

In a further embodiment, the inner and/or outer layers comprise between 50-100 wt% carbon, more preferably between 60-97 wt% carbon, and most preferably between 65- 95 wt% or between 70-95 wt% carbon, such as 70, 74, 80, 85, or 90 wt% carbon.

Binder

The first carbon particles of the outer layers may be bonded and shaped into a layer by a polymeric binder, such as polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof. Thus, the polymeric binder may constitute the remaining mass of the layer, e.g. in an embodiment of the disclosure, the outer layer comprises 10 wt% PVDF and 90 wt% carbon.

In an embodiment of the disclosure, the outer layers further comprise a polymeric binder, selected from the group of: polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof.

The inner layer preferably has the same properties as the outer layers, i.e. high SAC and electrical conductivity. Thus, the inner layer advantageously comprises the same carbon types, same carbon amounts, and same carbon particle sizes, and as the outer layers. Similarly, the carbon particles of the inner layer may be bonded and shaped into a layer by a polymeric binder.

In an embodiment of the disclosure, the carbon particles of the inner layer are similar to the carbon of the outer layers. In a further embodiment, the inner layer further comprises a polymeric binder, selected from the group of: polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof.

Cracks in the electrode will greatly increase the electrode resistance and thus the performance. In addition to the binders facilitating the shaping of the electrode film or layer, the presence of binders in certain amounts may also reduce the risk of crack formation as the mechanical strength and toughness may be improved due to the polymer. However, the capacitance of the electrode will typically also be reduced as the amount of binder content increases, since the pores are blocked by the polymers.

It was surprisingly found that the electrodes of the present disclosure facilitate a high amount of binder, while simultaneously facilitating high performance, as described in Example 5. It was surprisingly found that a binder content of above 20 wt%, such as 24 wt%, resulted in reduced crack formation in the electrode, and while simultaneously no significantly decrease in SAC or increased electrical resistance due to the larger binder content may be found, Specifically, for a binder content of 24 wt%, an in-plane electrical resistance of the dried slurry was measured to be as low as 0.6 mQ m, which is comparable to commercially available carbon fiber paper.

In an embodiment of the disclosure, the inner and/or outer layers comprise above 20 wt% polymeric binder, more preferably between 21-29 wt% polymeric binder, and most preferably between 22-28 wt%, such as 24, 25, or 26 wt%.

Carbon particle size

Since polymeric binders are generally not an electron conductor, the binder will reduce the electronic conductance between the carbon particles in the layers. As a result, the in-plane electron conductance of a layer may be low, even though the electronic conductance of the individual carbon particles is good.

To improve the electronic conductivity of a mixture of carbon particles and polymeric binder, the inner layer and/or outer layers advantageously comprise carbon particles with a range in average particle size, and preferably carbon particles with a bimodal size distribution, such that the particles comprises a fraction of larger particles and a fraction of smaller particles. For example, the carbon particles of each layer may comprise larger particles of e.g. activated carbon (AC) or graphite or carbon fibers having an average size between 50- 300 pm, such as a mesh size of 100 corresponding to a maximum particle size of 150 pm, and smaller particles of carbon black (CB) having an average size between 100 nm - 25 pm. The smaller CB particles will then fill the voids between the larger carbon (AC, graphite, carbon fiber) particles, similar to how sand can will fill up the voids between large rocks. By filling the voids using CB, a better conductive pathway between the larger carbon particles results, and also additional absorption surface area for ions is introduced. A distribution between the larger particles 2.4 and the smaller particles 2.5 resulting in improved electrical conductivity is illustrated in Figure 1B. Figure 7 shows a scanning electron microscopy image (SEI) of an embodiment of an electrode according to the present disclosure, comprising bigger particles of graphite and smaller particles of carbon black.

Further advantageously, the larger particles comprises carbon fibers. For example, to improve the environmental impact, the carbon fibers are advantageous recycled carbon fibers, such as milled carbon fiber powder from Easy Composites Ltd, as further described in Example 5. The carbon fibers advantageously have a typical or average fiber length of between 50-300 pm, more preferably between 60-200 pm, and most preferably between 75-150 pm, such as 80, 100, or 120 pm. Further advantageously, the carbon fibers have a typical or average fiber width of between 1-20 pm, more preferably between 3-15 pm, and most preferably between 5-10 pm, such as or 6.5,

7.5, or 8.5 pm. Further advantageously, the length and width ratio is selected to be between 100:1 to 100:20, such as 100:7.5. For example, advantageously the carbon fibers have a length of ca. 100 pm and a width of ca. 7.5 pm. The presence of carbon fibers, and specifically these dimensions of the carbon fibers, was seen to facilitate particularly high binder contents, while maintaining high electrode performance.

Figure 10 shows a scanning electron image of an embodiment of an electrode according to Example 5. It is seen that the carbon fibers tend to align in the plane of the electrode and create an electronically connected network. The carbon black particles fill up the voids that are left by the bigger particles to form a nanoporous material. Thus, the carbon fibers may promote further reduced pore diameters, thereby reducing the permeability, and making it harder for water to cross over from one channel to another. Hence, in this case, the carbon black particles perform three functions: they promote the hopping of electricity from one fiber to another, they absorb ions upon charging, and they limit the permeability of the electrodes for advective flows.

The carbon fibers may particularly facilitate a higher binder content and reduced permeability to the outer layers. For example, the dimensional configuration of the carbon fibers may prevent or reduce the amount of carbon fibers not present within the inner fibrous, scaffold support layer, as further exemplified in Example 5.

In an embodiment of the disclosure, the inner and/or outer layers comprise carbon particles with an average size between 100 nm - 300 pm, more preferably between 500 nm - 150 pm, and most preferably between 1 - 100 pm. In a further embodiment, the carbon particles have a bimodal size distribution comprising a fraction of larger particles, and a fraction of smaller particles. In a further embodiment, the bimodal size distribution has a first peak between 50-300 pm, and a second peak between 100 nm - 25 pm. In a further embodiment, the larger particles comprise activated carbon (AC), graphite, and/or carbon fibers, and the smaller particles comprise carbon black (CB).

By the term average particle size is meant the average particle size as determined by common particle size measurements techniques. The particle size of a spherical particle is unambiguously defined by its diameter or radius. However, for most cases, the particle shapes are not spherical, and for a batch of particles, e.g. a powder, the particles will differ in sizes and have a distribution of different sizes. Thus, when applying the common techniques as known to the skilled person for evaluating particle sizes, the particle size is often quantified in terms of a representative particle diameter or radius, such as the average particle diameter, or the fiber length.

The size of non-spherical particles may be quantified as the diameter of an equivalent sphere, such as the sphere having the same volume as the non-spherical particle, the sphere having the same surface area as the non-spherical particle, the sphere having the same sedimentation rate as the non-spherical particle, the sphere having a diameter corresponding to the length of the major axis (maximum length) of the non- spherical particle, or the sphere having a diameter corresponding to the minor axis (or minimum length) of the non-spherical particle. Despite this is not a proper quantification from a geometrical point of view, it is applied to provide a quantitative description of the characteristic sizes. In most cases, a particle size distribution exists. The average particle size may then refer to the particle diameter of the equivalent spherical particle as evaluated by laser diffraction. This particle diameter and the associated particle size distributions are evaluated using laser diffraction, where the liquidly dispersed particles are passed through a focused laser beam, such that the particles scatter the light. The scattering angle is proportional to the particle size, and a map of the scattering intensity versus the angle may then be obtained and used to calculate the particle sizes and the distribution. The calculation of the particle size distribution may be based on Mie theory, which is based on assuming spherical particles. The Mie theory includes comparison of the obtained scattering pattern with scattering patterns derived from theory (assuming spherical particles). Based on the particle size distribution, the average particle size may be determined.

The amount of the smaller particles compared to the larger particles also affects the resulting electronic conductivity of a layer, because it affects the carbon particle network which may be formed within each layer. Advantageously, the amount of smaller particles is such that the smaller particles essentially fill the void between the larger particles. Thus, the particles are abutting to form a closest packing, and the interstices between neighbouring particles form the porosity. Thus, a high amount of smaller particles may provide both high electron conductivity, as well as high surface areas and high volumetric SAC values.

In an embodiment of the disclosure, the carbon particles of the inner and/or outer layers comprise a fraction of between 20-80 wt% smaller particles, more preferably between 25-60 or 30-60 wt% smaller particles, such as 26, 28, 30, 32, 33, 35, or 40 wt%.

In a further embodiment, the carbon particles comprise a fraction of between 40-70 wt% larger particles, more preferably between 45-65 wt%, such as 48, 50, 55, 57, or 60 wt%.

To further improve the electrical conductivity as well as the SAC values of the layers, the smaller particles advantageously comprise a carbon material with higher ion absorbing properties, while the larger particles advantageously comprise a carbon material with higher conductivity, such as graphite and/or carbon fibers. Carbon materials with higher SAC or ion absorbing properties include activated carbon. The activated carbon has intrinsic porosity that enables adsorbtion, which is a result from the activation process. Carbon materials with higher SAC values also include carbon particles with high surface areas, because the ions can adsorb on the outer surface of the carbon particles due to the high surface energy. For example, for carbon particles with a specific surface area of above 100 m ² /g, the high surface area and related surface energy induce electrostatic interactions facilitating high ion adsorption. This surface adsorption or absorption capacity may be lower compared to the intrinsic capacity in AC, but it is faster because the ions do not need to travel inside the pores of the large activated carbon particles. Thus, advantageously, the smaller particles comprise AC or have a high surface area.

In an embodiment of the disclosure, the larger particles comprise activated carbon (AC), graphite, and/or carbon fibers, and the smaller particles comprise carbon black (CB).

Examples 3-6 describe further embodiments of electrodes comprising graphite or carbon fibers in combination with carbon black.

Outer layers

Advantageously, the outer layers comprise a mixture of carbon particles and polymeric binder, as well as the interstitial porosity. The porosity includes both the interstitial porosity between neighbouring particles, as well as the intrinsic or intra-particular porosity within the particles of e.g. AC. Hence, the distribution and packing of the carbon particles determine the porosity of the layer, in combination with the intrinsic porosity. The interstitial and intrinsic pores provide pathways for the adsorption of ions and migration of ions across the layer, as indicated by the coarse dashed lined arrow 6 in Figure 1A.

The outer layers therefore advantageously comprise a high porosity, and advantageously a high fraction of micro- and/or meso porosity, which facilitate ion adsorption and ion migration into the pores, e.g. due to the surface energy. However the pore size should not be too small as this hinders migration of ions into the pores. Moreover, the porosity should not be too high, at the expense of the mechanical stability and electron conductivity and risk of cross-flow between neighbouring channels. By the term micro porosity is meant pore diameters of less than 2 nm. By the term meso porosity is meant pore diameters between 2-50 nm. Preferably, the porosity of the outer layers constitutes of between 50-100 vol% micro- and/or meso porosity, more preferably between 70-90 vol%.

In an embodiment of the disclosure, the outer layers have a porosity of between 20- 60%, more preferably between 25-50%, and most preferably between 30-40%, such as 35%. In a further embodiment, the porosity constitutes of micro porosity and/or meso porosity.

The porosity and pore size distribution may be evaluated using measurement techniques such as mercury porosimetry or gas adsorption, also known as BET (Brunauer-Emmett-Teller) methods. Depending on the porosity of the layer, different methods may be preferred. In a preferred embodiment of the disclosure, the porosity and pore sizes are evaluated by BET based nitrogen adsorption.

Higher porosity implicitly reduces the electrical conductivity of a layer, because the pores are non-conducting. Also, the thicker the layer, the higher the electrical resistance of the layer. To optimize the SAC and at the same facilitate sufficient electrical conductivity, the outer layers advantageously comprise a porous network with low tortuosity, such as a tortuosity below 10, e.g. between 1-5, and the thickness of the outer layers are below 200, 150 or 100 pm, and a mass corresponding to a dry mass of between 100-300 g/m ² .

In an embodiment of the disclosure, the outer layers have a thickness of between IQ- 200 pm, more preferably 50-150 pm, and most preferably 70-120 pm, such as 100 pm.

In a further embodiment, the outer layers have a dry mass of between 100-300 g/m ² , more preferably between 150-250 g/m ² , such as 200 g/m ² .

Inner layer

The inner layer 2.2 further comprises an internal fibrous, scaffold support, as seen in the close up of the inner layer in Figure 1B. The fibers 2.3 are physically connected and thereby form a three-dimensional scaffold or network or backbone, which provides a mechanical framework within the inner layer. Advantageously, the microstructure of the inner layer may be identical to the microstructure of the outer layers, apart from the fibrous scaffold.

The fibrous scaffold is advantageously a self-supporting or free-standing structure, which may be manually and independently handled, before it is part of the inner layer. Hence, the fibrous scaffold may be a fibrous paper or felt, such as glass fiber paper or carbon fiber paper or felt. It follows that the thicker the paper, the higher the mechanical strength, and the more robust towards manually handling.

The fibrous scaffold will affect the conductivity of the structure. For example, if the fibrous scaffold comprises a material which is not an electron conductor, the scaffold is expected to reduce the electrical conductance between the carbon particles in the layer, in a similar manner as the polymeric binder. However, it was surprisingly found that a fibrous, scaffold support comprising non-conducting glass fiber paper, or low conducting carbon fiber paper or felt, resulted in surprisingly efficient electrodes which were also surprisingly mechanically strong. Due to the high efficiency, it was also surprisingly found that relatively thick, and correspondingly mechanically strong, fiber papers provided sufficient SAC and electronic conductivity. To improve the cost- efficiency of the electrode, the fibrous scaffold advantageously comprises or consists of glass fiber paper. Further it was found that glass fiber based electrodes facilitated an improved mechanical strength and toughness, as further described in Examples 1 and 6.

In an embodiment of the disclosure, the fibrous, scaffold support comprises glass fiber paper, and/or carbon fiber felt or paper. In a further embodiment, the fibrous, scaffold support consists of glass fiber paper. In a further embodiment of the disclosure, the inner layer has a thickness of between 200-900 pm, more preferably between 240-600 pm, and most preferably between 280-400, such as 300 pm.

The fibrous scaffold may reduce the porosity of the inner layer structure compared to the outer layers. The porosity of a commercial self-supporting and free-standing glass fiber paper may be approximately 93 vol%, corresponding to 7 vol% solid fiber glass. In an embodiment of the disclosure, the fibrous scaffold support has a porosity of between 80-97 vol%, more preferably between 85-95 vol%, such as ca. 93 vol%.

In addition to the fibrous scaffold, the inner layer advantageously comprises a mixture of carbon particles and polymeric binder. The presence of carbon particles and/or polymeric binder distributed within the glass fiber paper will further reduce the porosity.

Similar to the outer layers, the inner layer advantageously comprise a high porosity, and advantageously a high fraction of micro- and/or meso porosity, which facilitate ion adsorption and ion migration into the pores, e.g. due to the surface energy, and which do not hinder migration of ions into the pores. However, the porosity should not be too high, at the expense of the electron conductivity. Advantageously, the porosity of the inner layer is between 10-60%.

In an embodiment of the disclosure, the inner layer has a porosity of between 10-60%, more preferably between 20-50%, and most preferably between 25-40%, such as 35%. In a further embodiment, the porosity constitutes of micro porosity and/or meso porosity.

Similar as for the outer layers, the porosity and pore size distribution of the inner layer may be evaluated using measurement techniques such as mercury porosimetry or gas adsorption, also known as BET (Brunauer-Emmett-Teller) methods. Depending on the porosity of the layer, different methods may be preferred. In a preferred embodiment of the disclosure, the porosity and pore sizes are evaluated by BET based nitrogen adsorption. For example, the BET specific surface area and the pore size distribution are preferably analyzed using a Micromeritics Inc., TriStar II 3020.

Manufacture

The carbon particles of the inner layer and outer layers are advantageously identical, i.e. the first carbon particles of the outer layers are the same as the second carbon particles of the inner layer. This further has the advantage that the electrode may be manufactured by a simple process, e.g. by deposition and impregnation into a fibrous, scaffold support structure. Impregnation of a slurry requires a sufficiently flowable slurry. The flowability of the slurry into the pores will particularly depend on the homogenization, viscosity, temperature, stability, and rheology of the slurry. For example, a homogenized slurry with increased viscosity, and/or elevated temperature, and/or thixotropic properties, and/or absent isopropanol were found to have a flowability particularly suitable for impregnating a fibrous scaffold support, as described in Examples 3-4.

For example, a flowable slurry comprising the carbon particles may be applied or deposited onto the surfaces of a planar, fibrous, scaffold support. When the slurry is deposited onto the first surface of the fibrous, scaffold structure plane, the slurry forms a deposited layer on the first surface, corresponding to an outer layer. Since the slurry is further configured to be sufficiently flowable, i.e. to have a viscosity and surface tension/energy relative to the first surface, such that a part of the slurry will flow into the scaffold structure of the first surface, the scaffold is further partly or fully infiltrated or impregnated by the slurry. The infiltration of the scaffold structure with carbon particles will thereby form the inner layer, or the part of the inner layer facing the deposited layer. Subsequently, the slurry may be deposited onto the second surface of the scaffold, whereby the second outer layer may be formed, and the inner layer further formed by impregnation.

In an embodiment of the disclosure, the electrode is obtained by deposition and impregnation of a sufficiently flowable slurry comprising carbon particles onto at least two surfaces of the fibrous, scaffold support. In a further embodiment, the sufficient flowability is obtained by the slurry being thixotropic and/or having a sufficient viscosity and/or deagglomeration to facilitate the impregnation. It was found that a sufficiently flowable slurry, optionally due to high deagglomeration may be obtained by the electrodes comprising carbon fibers. Hence, a support entirely non-conductive across the thickness of the inner layer, may advantageously be impregnated, as further described in Example 5.

Thus, the layered electrode may be manufactured by a simple and non-complex process. To ensure sufficient flowability of the slurry, and sufficient infiltration into the inner layer, the slurry advantageously comprises an organic solvent, dispersants, and/or viscosity regulators, in addition to the polymeric binder and carbon powder. In an embodiment of the disclosure, the process for manufacturing an electrode, comprises the steps of: a) mixing at least one polymeric binder with an organic solvent, b) adding one or more dispersants and/or viscosity regulators, and carbon powder to the mixture, c) stirring the mixture to form a sufficiently flowable slurry, d) providing a fibrous, scaffold support, e) depositing a layer of the sufficiently flowable slurry on both sides of the porous support, thereby impregnating the fibrous, scaffold support, and f) drying the applied slurry to evaporate the organic solvent.

To ensure a deposited layer of sufficient thickness, the mixture advantageously comprises at least 10 wt% carbon powder. Alternatively, or additionally, the deposition may be repeated at least two times on each side of the support.

In an embodiment of the disclosure, the mixture comprises between 5-40 wt% carbon powder, more preferably between 10-30 wt%, such as 15, 20, or 25 wt% carbon powder. In a further embodiment, steps (e)-(f) are repeated at least 2 times, more preferably 3 times, and most preferably between 4-10 times, such as 5, 6, or 7 times.

To improve the environmental impact of the process, the slurry additives are advantageously low or non-toxic additives, such as PVDF binder, and acetone, isopropanol, or DMSO as solvent, dispersant and/or viscosity regulator. These are in contrast to toxic additives such as NMP. The organic solvents, dispersants, and viscosity regulators further have the advantage of being evaporated at low temperatures, such that they are removed at relatively mild drying conditions.

In an embodiment of the disclosure, the polymeric binder is selected from the group of: polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof. In a further embodiment, the organic solvent, dispersant, and/or viscosity regulator is selected from the group of: acetone, isopropanol, ethanol, methanol, hexane, toluene, DMSO, NMP, DMF, and combinations thereof, and preferably is DMSO in combination with acetone or isopropanol. However, the presense of isopropanol may be disadvantageous to the slurry stability, as further described in Example 4.

In an embodiment of the disclosure, the organic solvent, dispersant, and/or viscosity regulator do not comprise or consist of isopropanol.

To facilitate the flowability of the slurry, and hence including the infiltration process, the slurry may efficiently stirred, and applied by different deposition techniques. A specific degree of stirring and deagglomeration may affect the slurry flowability and capability of being impregnated into a fibrous support, as further described in Example 3. Improved deagglomeration and sufficient flowability may be obtained by using a homogenizer, such as an immersion blender or an industrial homogenizer. Also, specific deposition techniques may facilitate advantageously uniform and controlled thicknesses and impregnation degrees. For example, dip coating may provide particularly uniform and reproducible layer thicknesses.

In an embodiment of the disclosure, the stirring includes mechanical and/or ultrasonic stirring, such as a pulsed mode ultrasonic stirring or a homogenizer, such as an immersion blender.

In an embodiment of the disclosure, the flowable slurry is deposited by a method selected from the group of: painting, brushing, roll coating, spraying, dip coating, impregnation, and combinations thereof.

The final electrode may further be treated to improve the surface properties, as further described in Example 1. Advantageously, the dried electrode was surface treated with a hydrophilic solution to reduce the water-repelling properties (hydrophobicity) induced by e.g. the PVDF. For example, the final electrodes may be dipped or flushed in isopropanol, methanol, ethanol, or other organic solvents that mix with water. This will remove the air from the electrode, and facilitate the electrode wetting and acceptance of water afterwards.

In an embodiment of the disclosure, the method further comprises a step of dipping the dried electrode of step (f) in a hydrophilic solution, optionally comprising isopropanol, methanol, ethanol, and/or water. System

A CDI system may include multiple electrodes 2, assembled into a stack, as shown in Figure 2. The voltage to the electrodes may be applied via current collectors 1 and electrical wires 4, and electrical shorts between neighbouring electrodes are prevented by separating spacers 3. The spacers may serve the dual function of both preventing electrical shorts between neighbouring electrodes and enable feed flow.

The layered electrodes 2 of the present disclosure are free-standing units, which may be easily assembled into the stack, without additional further mechanical supports, such as conventional support materials in the form of metal foils.

The layered electrodes further has the advantage of having percolating porosity across the layered plane, and no external support structure. This means that the electrode is capable of exchanging ions with feed channels on either side of the electrode, which may be the case in systems including multiple flow channels. Such systems have a higher desalination capacity due to the higher amounts of water flow. Due to the high efficiency, such systems may further be particularly suitable for miniaturization, e.g. as auxiliary system units for larger scale cleaning devices.

An example of a system including multiple flow channels is shown in Figure 3. A system or cell including multiple flow channels are also referred to as a multi-channel CDI (MC-CDI). The system comprises six electrodes (131-136) and consequently five flow channels (151, 152) between the neighbouring electrodes. However, the system may include any number of electrodes, such as 3, 8, or 20, depending on the intended flow capacity of the system. The system may further include any size of the electrodes.

The MC-CDI operates by a charging scheme thereby enabling desalination of one channel into another channel. This means that the salt ions are directly separated out of the feed water. Thus, the water flows may be simple, and the CDI may further operate at low pressures and room temperature and with a limited applied voltage and electrode capacity, which is energetically favourable. Specifically, the MC-CDI may be energy efficient for desalination of low salinity source waters, such as brackish water. Further, the MC-CDI may achieve high water recovery, such as up to 90%, due to the system design. Specifically, the electrodes should not be over-charged meaning that the electrostatic potential between any pair of electrodes should not exceed 1.23 V after charging, which is the standard potential for unwanted water electrolysis. During charging, a larger potential than 1.23 V can be applied, as a large part of this voltage will be attributed to resistive losses within the cell.

The electrodes receive a voltage from a controller or field generator (161), and a raw feed inlet of water (10) is flowed into at least one, or e.g. every second flow channel in arrow direction L, as seen in Figure 3. The channels comprising raw feed water to be purified or desalinated may also be referred to as product channels (152). The channels that are not product channels may be referred to as reject channels or collecting channels (151). The collecting channels may be provided with a reservoir inlet (102) providing a fluid suitable for collecting the transferred ions. For example, the collecting channels may be supplied with a reservoir fluid suitable for dissolving the transported ions. The different channels may be connected and/or separated using e.g. a suitable manifold system.

When a voltage is applied, the salts absorb or desorb on the electrodes depending on their charge and the polarity of the voltage, and the electronic charge that is deposited or removed from the electrodes (electrons), which can be controlled by the polarity and strength of the voltage. Desorption can also occur by just lowering the voltage while keeping the polarity the same as during absorption. When the field generator is configured for generating a predetermined cycle of phases within an electric and/or magnetic field, e.g. phases were the different electrodes have different polarity, then the salts may be transported in the arrow direction I, and thus may be transported from a product channel, through the electrodes and into the neighbouring collecting channels (151). Hence, the generated field induce a current of ions in the channels that result in an exchange of ions with the electrodes.

Advantageously, the field generator may be configured to provide a variable field which, during the deionization of the feed liquid, varies in the channels over time in one or more of the group consisting of: polarity, strength, direction. For instance, the field may change polarity or reverse direction overtime according to a predetermined cycle, e.g. where a first field is generated between the channel electrode (131) and the separating electrode (132), and a second field is generated between the separating electrode (132) and the collector electrode (133), and wherein the first and second fields are configured to have a cycle of four phases, thereby generating an ion flow (I) from the process channel to the collector channel(s), wherein the ion flow direction is the same for anions and cations. This may for example be obtained by supplying the electrodes with a sine wave current and/or voltage, or a differently pulsed current and/or voltage, and thereby provide a variable field over time in each of the channels.

For example, Figure 8 shows an embodiment of a CDI device including a field generator configured for four phases. The device comprises four electrodes (denoted S _a , M _a , M _b , S _b ), providing a product channel between the M _a and M _b electrodes, and collecting channels between respectively the S _a and M _a electrodes, and between the M _b and S _b electrodes.

The field generator induces a first alternating electric current/field in at least the product channel, associated with flow of either cations or anions from the product channel, into and through the separating electrodes M _a and M _b . The field generator further induces a second alternating electric current/field in at least the two surrounding collector channels, associated with flow of either cations or anions into the collector channel and the collecting electrodes S _a and S _b . The field generator is further configured such that the alternating field of the product channel is delayed in phase with respect to the alternating field and current in the collector channels, e.g. delayed by less than half of a full cycle time. The delay in phase between the two fields may result in 4 distinct phases as sketched in Figure 8. This will result in that both cations and anions are moved from the product channel in phase 1, and finally into the collector channels in phase 4.

The field generator is electrically connected to the electrodes, e.g. in wired connection as illustrated in Figure 8, and the electric current directions or polarity of the electrodes are indicated by the arrows above the electrodes:

In phase 1, S _a and M _a are negatively polarised, and M _b and S _b are positively polarised. Hence, anions and cations are adsorbed into and through respectively M _a and M _b , and anions and cations are desorbed from respectively S _b and S _a into the collector channels, as indicated by the arrows.

In phase 2, S _a and M _b are negatively polarised, and M _a and S _b are positively polarised. Hence, the anions and cations are respectively adsorbed into S _a and S _b , as indicated by the arrows. In phase 3, M _b and S _b are negatively polarised, and S _a and M _a are positively polarised. Hence, anions and cations are adsorbed into and through respectively M _b and M _a , and anions and cations are desorbed from respectively S _a and S _b into the collector channels, as indicated by the arrows.

In phase 4, M _a and S _b are negatively polarised, and S _a and M _b are positively polarised. Hence, the anions and cations are respectively adsorbed into S _b and S _a .

Thus, the phases result in ions are moved from the product channel into the separating electrodes M _a and M _b , and subsequently move through and from the separating electrode into the collector channels. Thus, the field generator charges the M _a and M _b electrodes with either cations or anions from the product channel, and the phase- delayed second field discharges these ions again, and they end up in the collector channel.

Examples of field generated cycle of phases, and embodiments of CDI systems which may include one or more of the electrodes according to the present disclosure is disclosed in WO 2019/180151 , and the document is hereby incorporated by reference in its entirety. The electrodes according to the present disclosure may act as any electrode, e.g. as channel electrode (131), separating electrode (132), and/or collecting electrode (133), which separate any combination of neighboring product channels and/or neighboring collecting channels.

The electrodes and flow channels may be arranged in any configuration, and the flows controlled and distributed by flow controlling means, such as adapted by use of manifolds.

Figure 4 shows an embodiment of a system, where a raw feed inlet (10) of water is only flowed into the outermost left product channel (152). The adsorped ions are then transported in the arrow direction I, through the electrodes, the intermediate channels (153), and into the collecting channel (151), which is the outermost right channel. The ions transport may be driven by the voltages of each of the electrodes, as well as the concentration gradient across the intermediate channels, and between the outermost left and outermost right channel. To enable sufficient separation between flow channels, the electrodes should not have large voids that permit liquid leak through the electrodes. The transport across the electrode plane should primarily be transport of ions. Figure 6B shows an embodiment of the layered electrode according to the present disclosure. The electrode is exposed to a bright light source behind the electrode, and it is seen that no voids are present, because the light is not penetrating the electrode. In contrast, Figure 6A shows a commercial single layered carbon electrode based on aerogel, where three spots with light penetration is seen.

Examples of systems based on the electrodes according to the present disclosure are further described in Examples 1-7, and specifically Examples 1 , 3, and 7.

Examples

The invention is further described by the examples provided below.

Example 1 : Electrode manufacturing with graphite

Electrodes were produced using the following starting materials and procedure: 5.2 g. PVDF powder obtained from Ambofluor, Germany is mixed with 105 g. acetone (lab grade). While stirring, 51 g. DMSO is added. The PVDF will dissolve in a few minutes. Afterwards, graphite powder as well as carbon black (Vulcan XC-72, Cabot) are added in amounts of 30.8 g. and 17.3 g. respectively. The graphite powder may also be exchanged for activated carbon powder (DARCO, 100 mesh particle size). Thus, the weight ratio of the graphite:carbon black (Vulcan XC-72): PVDF is 58:32:10. The carbon black may have a bulk density of 96 kg/m ³ and an average particle size of 50 nm, and a specific surface area of 230 m ² /g.

Subsequently, 83 g. acetone and 17.4 g. isopropanol (lab grade) are added. The mixture is stirred vigorously overnight, and then sonicated while stirring using an ultrasonic probe in pulsed mode. Each batch of slurry is immediately applied after preparation. The fibrous support material is a regular glass fiber paper (50 g / m ² , Product #4979, Dana Lim A/S Denmark), which measures 0.3 mm in thickness before coating. The carbon slurry is applied to the material with a paint brush. Using a brush, four coating steps are implemented to reach the desired overall thickness of 0.5 mm. Each new coating is applied after waiting for one day to enable intermittent evaporation of solvents. The residual DMSO is removed in an oven at 150 °C for two hours after the last coating has dried. There are good indications that the reported times can be reduced significantly.

The final electrode measures approximately 15 g (corresponding to a dry mass of approximately 15 g/m ² or 200 g/m ² ) at a thickness of 0.5 mm. The resulting electrodes is flexible and shows no visible cracks. No visible holes can be observed when the electrode is held in front of a bright light source, such as a flash light, which is an important indication of good separation capabilities between flow channels.

Using a four-electrode setup to eliminate the effect of contact resistances, the sheet conductivity is measured for electrodes prepared with activated carbon, graphite based electrodes, and carbon felt based electrodes. A conductivity corresponding with approximately 25 ohm for electrodes prepared with activated carbon has been measured.

Preferably, the final electrode was further treated to improve the surface properties. Advantageously, the dried electrode was surface treated with a hydrophilic solution to reduce the water-repelling properties (hydrophobicity) induced by e.g. the PVDF. For example, the final electrodes may be dipped or flushed in isopropanol, methanol, ethanol, or other organic solvents that mix with water. This will remove the air from the electrode, and facilitate the electrode wetting and acceptance of water afterwards.

The layers and the final electrode microstructure is examined by scanning electron microscopy (SEM) as shown in Figure 7, and the porosity, conductivity, and SAC is further examined. Preferably, the porosity is examined by use of BET-based nitrogen adsorption.

In a comparative experiment, similar electrodes were manufactured on a carbon fiber paper instead of a glass fiber paper. The glass fiber paper was observed to have improved mechanical stability.

Figure 5 shows performance data for a four electrode system, where the electrodes comprised glass fiber paper with activated carbon and PVDF. The salt concentration in the product channel (152) shown as solid black curve, and the collecting channel (151) shown as stippled curve were measured over time. Figure 5A shows normal operation, where salt is transported from the product channel (152) to the collecting channel (151). The salt concentration is seen to decrease in the product channel as the salt ions are adsorbed and removed via the electrodes. A corresponding increase in the salt concentration of the collecting channel is observed as the adsorbed ions are transported into the channel.

Figure 5B shows operating the transport mechanism in reverse, and a similar decrease and increase of the salt concentration in the channels are observed.

The outlet flow of the product channels (152) and the outlet flow of the collecting channels (151) may be circulated back, e.g. circulated back to the feed inlet reservoir or the fluid reservoir for the collecting channels. This way, the salinity of one reservoir will continue to increase, while the other will continue to decrease. Alternatively, the outlet flows are not recirculated.

Experiments are performed using different number of electrodes, different sizes of electrodes, different flow rates for the flow channels, as well as electrodes based on graphite and/or activated carbon.

Example 2: Comparative tests

The mechanical strength, SAC values and performance of the electrodes of Examples 1, 3-6 are compared to conventional electrodes. Higher strengths may be obtained for comparable SAC and higher performance values.

Specifically, the performance of the electrodes of Examples 1, 3-6 comprising carbon black (CB), which may act as ion absorber, is compared to conventional electrodes comprising activated carbon (AC) instead of CB.

It was found that electrodes made with carbon black, e.g. Ketjenblack, as an absorber have a lower electrode salt absorbance capacity (SAC), but they were found to charge faster, likely due to the small particle size of the carbon black and the easily-accessible ion-absorption sites within the electrode. Therefore, the desalination cycles can be performed at shorter cycle times and at higher currents, thereby providing a more efficient desalination, as further described in Examples 1 and 3. Example 3: Electrode manufacturing with graphite

The electrodes were made by a similar method as in Example 1, except the pulsed sonicator is replaced by a homogenizer, such as an immersion blender or an industrial homogenizer. Similar to Example 1, the weight ratio of the graphite:carbon black (Vulcan XC-72):PVDF is approximately 57:33:10, and the final electrode has a dry mass of approximately 200 g/m ² .

The homogenization resulted in a faster, scalable, and more reliable mixing process to form a sufficiently flowable and reproducible slurry. Specifically, the homogenized slurry was observed to have increased viscosity, and to be warm and thixotropic.

These properties were found to provide a slurry configured with a viscosity and flowability particularly suitable for impregnating a fibrous scaffold support. The thixotropic property is assumed to be related to a more efficient deagglomeration of the carbon black particle clusters.

Figure 9 shows experimental data for the salt concentration in the channels during operation for electrodes according to Example 3. For example the electrodes may be operated as described in Examples 1 and 7.

Example 4: Electrode manufacturing with isopropanol

The electrodes were made by a similar method as in Examples 1 and 3, except the isopropanol was not included or substituted with a different organic solvent, such as acetone. Slurries with improved life time in terms of stability, separation, viscosity, and sufficiently flowability particularly suitable for impregnating a fibrous scaffold support, were obtained.

Example 5: Electrode manufacturing with carbon fibers

The electrodes were made by a similar method as in Examples 1 , 3 and 4, except the graphite (i.e. the larger carbon particles) was replaced with carbon fibers. The carbon fibers are advantageous recycled carbon fibers, such as milled carbon fiber powder from Easy Composites Ltd.

The weight ratio of the carbon fibercarbon black (Vulcan XC-72):PVDF is approximately 48:28:24, It was surprisingly found that a binder content of above 20 wt%, such as 24 wt%, in combination with the carbon fiber resulted in reduced crack formation in the electrode, and simultaneously no significantly decrease in SAC or increased electrical resistance due to the larger binder content, Specifically, for a binder content of 24 wt%, an in-plane electrical resistance of the dried slurry was measured to be as low as 0.6 mQ m, which is comparable to commercially available carbon fiber paper.

The improved combination of reduced cracks, SAC and resistance may be due to the specific structure obtainable using carbon fibers. However, the improvement may also partly emerge from the improved outer layers of the electrode obtained during the impregnation of the glass fiber support.

Figure 10 shows a scanning electron image of an embodiment of an electrode according to Example 5. It is seen that the carbon fibers tend to align in the plane of the electrode and create an electronically connected network. The carbon black particles fill up the voids that are left by the bigger particles to form a nanoporous material. Thus, the carbon fibers may promote further reduced pore diameters, thereby reducing the permeability, and making it harder for water to cross over from one channel to another. In this case, the carbon black particles perform three functions: they promote the hopping of electricity from one fiber to another, they absorb ions upon charging, and they limit the permeability of the electrodes for advective flows.

In a further example, the carbon fibers are having dimensions of 100 pm length, and 7.5 pm width. Hence, though the slurry is configured to be sufficiently flowable to impregnate the fibrous, scaffold support,

Example 6: Electrode manufacturing without a fibrous, scaffold support The electrodes were made by a similar method as in Examples 1, 3-5, except the slurry was not applied on a fibrous, scaffold support, but shaped into a freestanding electrode, e.g. by casting.

The resulting electrodes were less reproducible due to the different slurry deposition, and the electrodes were seen to be less strong and less stable during operation, and prone to swelling and deformation. Example 7: Cell for multi-channel capacitive deionization (CPI)

The electrodes of Example 3 were tested in a cell for multi-channel CDI, and the results are shown in Figure 9.

Examples of multi-channel CDI is shown in Figures 3-4. In the present Example, a multi-channel CDI with 8 electrodes were tested. The cell consist of end electrodes Sa and Sb, as well as three pairs of middle electrodes Ma and Mb, in a configuration Sa,Ma1,Mb1,Ma2,Mb2,Ma3,Mb3,Sb.

The currents supplied to the electrodes are shown in Figure 9B, where the current to Ma1, Ma2, and Ma3 is the same and indicated by Ma, and the current supplied to Mb1, Mb2, and Mb3 is the same and indicated by Mb.

The experiment started with the pump pushing water into the cell at approx. 80 ml/min, and the electrical current was initially off. Between approximately 100 and 4000 seconds in the experiment, the 4-phase cycle (as described in Figure 8) was turned on, yet it was deliberately performed very fast (cycle time 200s) at relatively high currents. Despite the increased currents, the (middle) electrodes cannot charge sufficiently in this short time and no transport of salt is observed, as the conductivities in Figure 9A show.

When the cycle time of the 4-phase cycle is reduced to a cycle time of 1800 seconds at 4000 seconds into the experiment, as Figure 9B shows, the electrodes charge sufficiently and it can be observed in Figure 9A that the ions are transported between the channels.

The optimal specific cycle times will depend on the electrode properties. For example, electrodes with a higher dry mass load will generally require longer cycle times than electrodes with lower dry mass.

Advantageously, the 4-phase cycle of Figure 8 may include a more gradual transition between the phases, as can be seen from the fluid transition between the phases in Figure 9. Further advantageously, the field sizes and the electrodes may be configured for the location within the cell. In phase 2 and 4 ions are pulled out from the middle electrodes and into the collecting channels. However, the field generated by the outer electrodes may also be configured to pull the ions through the middle electrodes. The latter is enabled by the middle electrodes acting as a filter for one of the ion species (cations or anions) because they are loaded by either cations or anions at the end of phase 1 or 3. The relatively high concentration of one of these ion species creates a low-resistance pathway for those ions. Thus, in phase 2 and 4 the field needs not to be configured for the middle electrodes to discharge, and instead the field can be configured such that the current is running until the outer electrodes are saturated.

Hence, advantageously, the outer electrodes are optimized for providing high currents and high salt absorption by making them relatively thick and heavy (i.e. having a high dry load, e.g. above 200 g/m ² ). At the same time, the middle electrodes may be optimized to be light (i.e. having dry loads below 200 g/m ² ), such that they can charge quickly and rapidly enter a filtering state, which minimizes their electricity consumption.

Reference numbers

1 - Current collector

2 - Electrode

2.1 - Outer layer

2.2 - Inner layer

2.3 - Support fiber

2.4 - Larger particle

2.5 - Smaller particle

3 - Spacer

4 - Electrical wire

6 - Pathway for ions

7 - Pathway for electrons 10 - Raw feed inlet

102 - Reservoir inlet 131-136 - Electrode

151 - Collecting channel

152 - Product channel

153 - Intermediate channel References

[1] WO 2009/065023 Items

The presently disclosed may be described in further detail with reference to the following items.

1. An electrode, comprising an inner layer sandwiched between two outer layers, wherein the outer layers comprise multiple first carbon particles, and wherein the inner layer comprises an internal fibrous, scaffold support and multiple second carbon particles distributed within the scaffold support, thereby forming a three-dimensional, percolating network of second carbon particles. 2. The electrode according to item 1 , wherein the electrode is a capacitive deionization (CDI) electrode for electrosorption.

3. The electrode according to any of the preceding items, wherein the first and second carbon particles are the same.

4. The electrode according to any of the preceding items, wherein the first and/or second carbon particles are selected from the group of: activated carbon (AC), carbon black (CB), graphite, carbon fibers, and combinations thereof.

5. The electrode according to any of the preceding items, wherein the inner and/or outer layers comprise between 50 - 100 wt% carbon, more preferably between 60-97 wt% carbon, and most preferably between 65-95 wt% carbon, such as 70, 74, 80, 85, or 90 wt% carbon.

6. The electrode according to any of the preceding items, wherein the inner and/or outer layers further comprise a polymeric binder, selected from the group of: polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof.

7. The electrode according to item 6, wherein the inner and/or outer layers comprise above 20 wt% polymeric binder, more preferably between 21-29 wt% polymeric binder, and most preferably between 22-28 wt%, such as 24, 25, or 26 wt%.

8. The electrode according to any of the preceding items, wherein the inner and/or outer layers comprise carbon particles with an average size between 100 nm - 300 pm, more preferably between 500 nm - 150 pm, and most preferably between 1 - 100 pm.

9. The electrode according to any of the preceding items, wherein the inner and/or outer layers comprise carbon particles with a bimodal size distribution comprising a fraction of larger particles, and a fraction of smaller particles.

10. The electrode according to item 8, wherein the bimodal size distribution has a first peak between 50-300 pm, and a second peak between 100 nm - 25 pm.

11. The electrode according to any of items 8-9, wherein the larger particles comprise activated carbon (AC), graphite and/or carbon fibers, and the smaller particles comprise carbon black (CB).

12. The electrode according to any of items 9-11, wherein the carbon particles comprise a fraction of between 20-80 wt% smaller particles, more preferably between 25-60 wt% smaller particles, such as 26, 28, 30, 32, 33, 35, or 40 wt%.

13. The electrode according to any of items 9-12, wherein the carbon particles comprise a fraction of between 40-70 wt% larger particles, more preferably between 45-65 wt%, such as 48, 50, 55, 57, or 60 wt%.

14. The electrode according to any of the preceding items, wherein the outer layers have a porosity of between 20-60%, more preferably between 25-50%, and most preferably between 30-40%, such as 35%.

15. The electrode according to item 14, wherein the porosity constitutes of micro- and/or meso porosity.

16. The electrode according to any of the preceding items, wherein the outer layers have a thickness of between 10-200 pm, more preferably 50-150 pm, and most preferably 70-120 pm, such as 100 pm. 17. The electrode according to any of the preceding items, wherein the outer layers have a dry mass of between 100-300 g/m ² , more preferably between 150-250 g/m ² , such as 200 g/m ² .

18. The electrode according to any of the preceding items, wherein the fibrous, scaffold support comprises glass fiber paper, and/or carbon fiber felt or paper.

19. The electrode according to item 18, wherein the fibrous, scaffold support consists of glass fiber paper.

20. The electrode according to any of the preceding items, wherein the inner layer has a thickness of between 200-900 pm, more preferably between 240-600 pm, and most preferably between 280-400, such as 300 pm.

21. The electrode according to any of the preceding items, wherein the fibrous, scaffold support has a porosity of between 80-97 vol%, more preferably between 85-95 vol%, such as ca. 93 vol%.

22. The electrode according to any of the preceding items, wherein the inner layer has a porosity of between 10-60%, more preferably between 20-50%, and most preferably between 25-40%, such as 35%.

23. The electrode according to any of the preceding items, obtained by deposition and impregnation of a sufficiently flowable slurry comprising carbon particles onto at least two surfaces of the fibrous, scaffold support.

24. A process for manufacturing an electrode, comprising the steps of: a) mixing at least one polymeric binder with an organic solvent, b) adding one or more dispersants and/or viscosity regulators, and carbon powder to the mixture, c) stirring the mixture to form a sufficiently flowable slurry, d) providing a fibrous, scaffold support, e) depositing a layer of the sufficiently flowable slurry on both sides of the porous support, thereby impregnating the fibrous, scaffold support, and f) drying the applied slurry to evaporate the organic solvent. 25. The process according to item 24, wherein the mixture comprises between 5-40 wt% carbon powder, more preferably between 10-30 wt%, such as 15, 20, or 25 wt% carbon powder.

26. The process according to any of items 24-25, wherein the polymeric binder is selected from the group of: polyvinylidene fluoride (PVDF), polytetrafluoroethylene (PTFE), polyvinyl alcohol (PVA), and combinations thereof.

27. The process according to any of items 24-26, wherein the organic solvent, dispersant, and/or viscosity regulator is selected from the group of: acetone, isopropanol, ethanol, methanol, hexane, toluene, DMSO, NMP, DMF, and combinations thereof, and preferably is DMSO in combination with acetone or isopropanol.

28. The process according to item 27, wherein the organic solvent is not isopropanol.

29. The process according to any of items 24-28, wherein the stirring includes mechanical and/or ultrasonic stirring, such as a pulsed mode ultrasonic stirring or a homogenizer, such as an immersion blender.

30. The process according to any of items 24-29, wherein the sufficiently flowable slurry is deposited by a method selected from the group of: painting, brushing, roll coating, spraying, dip coating, impregnation, and combinations thereof.

31. The process according to any of items 24-30, wherein steps (e)-(f) are repeated at least 2 times, more preferably 3 times, and most preferably between 4-10 times, such as 5, 6, or 7 times.

32. The process according to any of items 24-31, further comprising a step of dipping the dried electrode of step (f) in a hydrophilic solution, optionally comprising isopropanol, methanol, ethanol, and/or water.