Membranes for RED

In 2007, Turek [32] studied the effect of the solution velocity on cell power output and process economy and observed that the main bottleneck for successful market introduction of RED is the membrane price. Nevertheless, most of the earlier work was dedicated to stack design and the effect of solution flow and composition, but not to ion exchange membrane characterization and performance testing. Ion exchange membranes are membranes with fixed anionic or cationic exchange groups that are able to transport cations or anions. The presence of these charged groups gives these membranes their specific properties and amount, type, and distribution of the ion exchange groups determine the overall membrane properties. Based on the type of fixed charge groups, ion exchange membranes can be classified as strong acidic and strong basic, or weak acidic and weak basic membranes. In strong acidic CEMs, sulfon groups serve as the fixed charged group in the membrane. Weak acidic membranes contain carboxylic acid as the fixed charged group. Quaternary and tertiary amines, respectively, provide the fixed positive charged groups in strong and weak basic AEMs (Fig. 9).

Two different types of ion exchange membranes can be distinguished, a classification that is based on the structure of the membrane: homogeneous and heterogeneous membranes. In homogenous ion exchange membranes, the fixed charge groups are evenly distributed over the entire membrane matrix. Homogenous membranes can be manufactured by polymerization

Figure 9 Typical example of (a) a cation exchange membrane (CEM) with SO3 groups as the cation exchange group and (b) an anion exchange membrane (AEM) with N(CH3)^ as the typical anion exchange group.

and polycondensation of functional monomers (e.g., fenylosulfonic acid with formaldehyde) or functionalization by, for example, postsulfonation [40-43]. Heterogeneous membranes have distinct macroscopic charged domains ofion exchange resins in a basically uncharged polymer membrane matrix. These membranes are usually produced by melting and pressing a dry ion exchange resin with a granulated polymer (e.g., polyvinylchloride) [44] or by dispersing the ion exchange resin in a polymer solution [45]. The distinct difference in structure between homogenous and heterogeneous ion exchange membranes also influences the properties of the specific membrane, as will be shown later.

Ion exchange membranes are the key elements in RED and the electrical resistance of the membrane and its permselectivity (the ability of the membrane to discriminate between cations and anions) are the most important membrane properties for RED because these properties directly influence the overall RED performance and power output. Because these properties are directly determined by the number of fixed charges inside the ion exchange membrane, the ion exchange capacity (IEC), the swelling degree (SD), and the fixed charge density of a membrane also play a crucial role.

The IEC [expressed in milliequivalent of fixed groups per gram of dry membrane (meq/g membrane)] is the number of fixed charges inside the ion exchange membrane per unit weight of dry polymer. The fixed charge density, expressed in milliequivalent of fixed groups per volume of water in the membrane (meq/L), is determined by this IEC and the SD of the membrane. The fixed charge density is lower in the swollen state than in the dry state because the distance between the charged groups is increased upon swelling of the membrane, while the number of charged groups remains unchanged. The concentration and the type of these fixed charged groups determine the electrical resistance and the permselectivity of the membrane, and these properties are directly related to the maximum power output obtainable in RED.

When an ion exchange membrane is in contact with an electrolyte (salt solution), ions with the same charge as the fixed charges in the membrane (co-ions) are excluded and cannot pass through the membrane, while the oppositely charged ions (counterions) can freely move through the membrane. This effect is known as Donnan exclusion [46]. Ion exchange membranes are never 100% selective and the permselectivity of an ion exchange membrane quantifies the ability of that membrane to discriminate between co-ions and the oppositely charged counterions.

Although the charge density has a strong influence on both the permselectivity and the membrane resistance, a straightforward relationship between the permselectivity and the membrane resistance does not exist as can be seen in Fig. 10 [32] (values for both AEMs and CEMs and homogeneous and heterogeneous membranes are shown).

In general, the resistance of heterogeneous ion exchange membranes is significantly higher than that of the homogenous types. This phenomenon can be related to the structure of the heterogeneous membranes: heterogeneous ion exchange membranes have distinct macroscopic domains of ion exchange resins in an uncharged polymer matrix. Consequently, the resistance of these heterogeneous membranes is higher. In general, less selective membranes have a lower membrane resistance than more selective ones, although this is only a general trend and several exceptions exist. In general, the permselectivity of CEMs is higher than the corresponding

Figure 10 Membrane permselectivity as a function of the membrane resistance (at 25°C). CEM is a cation exchange membrane (■) and AEM is an anion exchange membrane (□) [34].

values for AEMs. This is mainly due to the higher SD of AEMs, which reduces the effective fixed charge density and thus reduces the permselectivity.

Audinos [30], who was one of the first who systematically investigated the effect of two types of anion and CEM pairs on the power output in RED, already mentioned explicitly the importance of membranes specially developed for RED. Nevertheless, mainly due to limitations in availability of such membranes, most scientists use the above-presented standard electrodialysis membranes to study the performance of a RED system [30,31,38,47]. The manufacturer data available for these membranes do not offer sufficient information on the membrane properties relevant for RED and do not always allow mutual comparison of the different commercially available membranes, because of the different conditions often used for membrane characterization. Dlugolecki et al. [34] made a comprehensive overview of membrane benchmarking for RED. They experimentally determined a range of membrane properties of commercially available membranes relevant for RED under equivalent conditions to enable a fair comparison of the results and a proper evaluation of the different membranes for application in RED. Table 4 shows the experimentally determined values of these properties [34]. For comparison, the data of the manufacturers are also presented, although they are not always determined under equal conditions [48-51].

Table 4 clearly shows that the membrane characteristics vary over a wide range and strongly depend on the type of membrane and the differences in molecular structure and composition of the membranes. In general, the data provided by the manufacturers are in reasonable good agreement with the experimentally determined values, with some exceptions.

The IEC presented in Table 4 represent the number of strong acidic (—SO—) groups in CEMs and strong basic (-NR|) groups in the AEMs. Although the experimentally determined IEC is generally in good agreement with the data supplied by the manufacturers, strong deviations are visible for the APS membranes of Selemion and the FAD membranes from Fumasep. Both AEMs consist of a mixture of weak and strong ion exchange groups, but the experimental method used to determine the IEC only allows the detection of strong basic groups, whereas weak basic groups are not recognized. This results in significantly lower experimental values for the IEC, compared to the manufacturer's data. In general, SD values are similar to the data of the manufacturers, although the experimentally determined SD of the Selemion APS membrane is extremely high, which is probably due to the rough membrane surface of the APS membrane, which affects the wiping off of water from the membrane surface before measuring the weight. The thickness of the membrane strongly depends on the type of the membrane: Homogenous membranes are generally thinner than heterogeneous membranes, which is due to the structure of the membrane and its preparation method [40-43,52,53].

Based on these experimentally determined data, Dlugolecki et al. [34] applied a theoretical model to evaluate these specific membrane properties in relation to the expected performance of these membranes under RED conditions [28,34]. This model relates the membrane resistance (R,em and Rcem) and its permselectivity (aav) directly to the maximum power output in RED [Wmax (W)]. Membrane resistance and membrane permselectivity are the two most important parameters in this respect because they indirectly also include the membrane thickness and structure, its IEC and SD, and thus the fixed charge density:

Table 4 Experimentally determined membrane characteristics of several commercially available ion exchange membranes (bold) [34]

Membrane IEC (meq/g Permselectivitya Resistance13 SD (%) Thickness Properties dry) (%) (O.cm2) (mm)

Cation exchange membranes



1.36 >1.0









Electrolysis, high selectivity


1.14 >1.0









Diffusion dialysis for NaOH



2.30 2.0-2.5









Low electric resistance


1.62 1.5-1.8









High mechanical strength

Ralex® (Heterogeneous)


2.34 2.2






< 55


< 700

Electrodialysis, Electrodeioniza-tion



2.01 N/A










Anion exchange membranes













Diffusion dialysis for acid













Low electric resistance












Resistant against organic fouling












High mechanical strength

Ralex® (Heterogeneous)












Electrodialysis, Electro deionization













Diffusion dialysis, low resistance












Diffusion dialysis, oxidant proof

Note: For comparison the data given by the membrane manufacturers are also presented [48—51]. a Membrane potential measured across the membrane between 0.5 and 0.1 M solutions. b Measured in 0.5M NaCl solution at 25°C. c Measured by electrophoresis, 2mA/cm2.

d Determined by 1 kHz AC measurement in the 0.5 N NaCl solution at 25°C.

where N is the number of membrane pairs (one cell pair consist of one anion and one CEM), aav the average membrane pair permselectivity (—), R the universal gas constant [8.314J/(molK)], T the absolute temperature (K), F the Faraday constant (96,485 C/mol), ac the concentrated solution activity (mol/L), ad the diluted solution activity (mol/L), Raem the AEM resistance (O m2), Rcem the CEM resistance (O m2), A the effective membrane area (m2), dc the thickness of the concentrated compartment (m), dd the thickness of the diluted compartment (m), kc the concentrated compartment conductivity (S/m), and kd the diluted compartment conductivity (S/m).

In order to compare commercially available membranes with each other, it is more convenient to convert the power output into power density, which is the power output normalized for the membrane area (W/m2):

W max

where Pmax is the maximum power density (W/m2), Wmax maximum power output (W), A the effective membrane area (m2), and Nm the number of membranes (—).

As Eqs. (23) and (24) predict the theoretical power output of the total system under RED conditions in relation to the individual membrane characteristics, it can be used as a tool to evaluate and compare the different anion and cation exchange membranes with respect to their performance in RED. Dlugolecki et al. [34] evaluated the relative importance of membrane resistance and permselectivity on the power density in a RED stack. Fig. 11 shows the power density as a function of the membrane resistance and permselectivity for two different spacer thicknesses (a) 600 and (b) 150 mm.

When thicker spacers (>600 mm, Fig. 11a) are used in the system, the power density of the system is hardly dependent on the membrane resistance or permselectivity. In this case, the resistance of the dilute compartment dominates the overall process resistance and maximum power densities of only 2 W/m2 can be obtained. When the distance between the membranes is decreased (Fig. 11b), the effect of the membrane properties and thus the difference in power density of the different membranes becomes more pronounced. With increasing permselectivity and especially decreasing membrane resistance, the power density significantly increases and values as high as 7 W/m can be obtained with properly designed stacks. Nevertheless, the process requires a minimum in spacer thickness

Figure 11 Relationship between the power density, the membrane permselectivity, and the membrane cell pair resistance for membrane pair with (a) 600 mm and (b) 150 mm thick spacers. Model calculations are based on seawater (0.5 M NaCl) as concentrated salt solution and river water (0.05 M NaCl) as diluted stream (T = 25 °C) [34].

Figure 11 Relationship between the power density, the membrane permselectivity, and the membrane cell pair resistance for membrane pair with (a) 600 mm and (b) 150 mm thick spacers. Model calculations are based on seawater (0.5 M NaCl) as concentrated salt solution and river water (0.05 M NaCl) as diluted stream (T = 25 °C) [34].

because at too thin spacer thicknesses, the energy consumption for solution pumping increases tremendously due to the high pressure drop over the compartments.

Because Eqs. (23) and (24) can also be used to predict the performance of only a cation or only an AEM in RED, Dlugolecki et al. [34] used their experimental data presented in Table 4 as input values for the model calculations to predict the maximum power density obtainable with each specific membrane (Fig. 12a and b). In this case, the average membrane pair selectivity (aav) is replaced by the individual membrane selectivity of the cation or anion exchange membrane, respectively, whereas in the case of a CEM the corresponding resistance of the AEM is neglected, and vice versa when only an AEM is used. The thickness of the concentrated and diluted compartment is divided by a factor 2. Dlugolecki et al. assumed that seawater has a NaCl concentration of 0.5 M (g± — 0.686 and kc — 4.648 S/m, T — 25 °C) and river water has a concentration of 0.05 M NaCl (g + — 0.820 and kc — 0.551 S/m, T — 25 °C).

It is obvious that the power density strongly depends on the spacer thickness (as presented before) and also on the type of membrane. The resistance of the heterogeneous membranes investigated is too high to be useful in RED. Even in a perfectly designed RED stack (extremely thin spacers), it is not possible to obtain power densities higher than 1.5 W/m2.


0 150 300 450 600 750 900 Spacer thickness (|jm)

0 150 300 450 600 750 900 Spacer thickness (jim)



0 150 300 450 600 750 900 Spacer thickness (|jm)

0 150 300 450 600 750 900 Spacer thickness (jim)

Figure 12 Prediction of the maximum obtainable power density based on experimental membrane characterization for (a) anion exchange membranes and (b) cation exchange membranes [34].

Homogeneous membranes are more suitable for RED. Based on these results, the best benchmarked AEMs are Neosepta AFN from Tokuyama Co. (Japan) and Selemion APS from Asahi Glass Co. Ltd. (Japan), with a predicted power density of more than 5 W/m2 (at a spacer thickness of 150 mm). The Neosepta CM-1 CEM from Tokuyama Co. (Japan) shows the best performance as CEM for RED and reaches a theoretical power density of more than 4W/m2.

Although this model is a very useful tool to make a rough estimation of the performance of the different membranes under RED conditions, it is a theoretical model that includes several assumptions [34]: (i) concentration polarization phenomena near the membrane surface are negligible due to the small current densities obtained through the membranes and (ii) the resistance of the electrodes is assumed to be negligible compared to the membrane resistance. This assumption is allowed when the resistance of the membranes is large compared to the resistance of the electrodes, which can be obtained when a large number of membrane cell-pairs is used (as will be required anyway to generate sufficient power at low costs), and (iii) the feed solution does not change in concentration along the channels. This assumption has a strong relationship with the feed channel design. Although assumptions (i) and (iii) are valid assumptions for a first initial comparison under laboratory conditions, they will become an important issue in the real application where real river and seawater are used.

Was this article helpful?

0 0
Waste Management And Control

Waste Management And Control

Get All The Support And Guidance You Need To Be A Success At Understanding Waste Management. This Book Is One Of The Most Valuable Resources In The World When It Comes To The Truth about Environment, Waste and Landfills.

Get My Free Ebook

Post a comment